A variety of sophisticated applications are currently utilizing artificial intelligence (AI), but the results of many AI models are difficult to understand and trust due to their black box nature. In addition, novel regulations and highly regulated areas have made the auditability and verifiability of decisions obligatory, creating an increased demand for the ability to understand such applications. It is therefore imperative that AI decisions are trustworthy, transparent, and comprehensible to humans, and there is a need for eXplainable AI (XAI) methods to improve trust in AI models. Therefore, XAI has become a popular research topic in the field of AI.

Literatur

• Adadi, A., & Berrada, M. (2018). Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI). IEEE Access, 6, 52138-52160.
• Carvalho, D.V., Pereira, E.M., & Cardoso, J.S. (2019). Machine Learning Interpretability: A Survey on Methods and Metrics. Electronics, 8, 832.

Liste der möglichen konkreten Themen:

• AI-BA-1-1, Sommersemester 2026, Betreuung: M.Sc. Luca Gemballa

  Explaining AI with AI: Using LLMs to Explain Medical Machine Learning Models

  Although healthcare constitutes one of the focal domains for XAI use, many prominent XAI methods are not intuitive to interpret for most clinicians. While counterfactual explanations come closer to human explanatory intuition compared to more abstract approaches like feature attribution, they still lack some essential qualities to provide suitable human-centered explanations. Large Language Models (LLM) may offer a solution for this inadequacy. By generating text-based explanations either based on the ML model directly or with respect to some other explanation technique, LLMs can provide more actionable, clinically-relevant, and accessible explanations.

  In the scope of this seminar paper, a literature review will be conducted to explore how LLMs are applied as means of explanation in medical AI systems. 

  Literatur

  • Lee, S., Cho, W. I., Lee, Y., Kim, D. J., Nam, K. H., Lee, S., Suh, J & Ko, T. (2025). A prompt framework for enhancing LLM-based explainability of medical machine learning models: an intensive care unit application. BMC Medical Informatics and Decision Making, 25(1), 430.
  • Michalowski, M., Wilk, S., Bauer, J. M., Carrier, M., Delluc, A., Le Gal, G., Wang, T.-F, Siegal, D. & Michalowski, W. (2024). Manually-curated versus LLM-generated explanations for complex patient cases: an exploratory study with physicians. In International Conference on Artificial Intelligence in Medicine (pp. 313-323). 

• AI-BA-1-2, Sommersemester 2026, Betreuung: M.Sc. Luca Gemballa

  Using Retrieval Augmented Generation for Explainability

  Some of the fundamental goals of XAI are to make AI systems more transparent, understandable, and trustworthy. While a substantial part of the XAI community has focused on developing and evaluating methods for creating feature attribution values or counterfactuals, other paths to enhance model understanding have seen less development. With the recent advances in generative AI systems however, researchers are increasingly proposing XAI methods involving natural language. Retrieval Augmented Generation (RAG) is an essential aspect in this endeavor, as it can relate generated outputs to a database of prespecified, relevant documents that serve as context and ensure trustworthy results.

  In the scope of this seminar paper, a literature review will be conducted to explore how RAG is applied to enhance user understanding, how it is integrated in user applications, and which domains are at the forefront of applying RAG for explainability. 

  Literatur

  • Pujari, T., Pakina, A. K., & Goel, A. (2023). Explainable AI and governance: Enhancing transparency and policy frameworks through retrieval-augmented generation (RAG). IOSR Journal of Computer Engineering, 25(6), 65-79.
  • Ranaldi, L., Valentino, M., & Freitas, A. (2025, April). Eliciting critical reasoning in retrieval-augmented generation via contrastive explanations. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (pp. 11168-11183). 

• AI-BA-1-3, Sommersemester 2026, Betreuung: M.Sc. Luca Gemballa

  Beyond the usual XAI Plots with Visual Analytics

  The field of XAI has tried and tested numerous ways of deriving and presenting explanations. Visualizations are a common aspect in this, as methods like feature attribution values, heatmaps, or prototypes are commonly or even exclusively presented in graphical form. Beyond these “standard” XAI visualization, other plots exist that can present information in order to explain AI outputs. The domain of visual analytics is adjacent to XAI and offers a rich literature on different visualization types that can be used to complement established XAI visualizations. We are interested in how visual analytics approaches have attempted to raise user trust, understanding, and transparency, how different visualizations have been combined with each other and with other modalities like texts or tables, and which effects are associated with these approaches. 

  In the scope of this seminar paper, a literature review will be conducted to explore the effects of different visual analytics approaches on users and their combination with XAI methods. 

  Literatur

  • Alicioglu, G., & Sun, B. (2022). A survey of visual analytics for explainable artificial intelligence methods. Computers & Graphics, 102, 502-520.
  • Laguna, S., Heidenreich, J. N., Sun, J., Cetin, N., Al-Hazwani, I., Schlegel, U., Cheng, F. & El-Assady, M. (2023). ExpLIMEable: A visual analytics approach for exploring LIME. In 2023 Workshop on Visual Analytics in Healthcare (VAHC) (pp. 27-33). 

Cognitive and emotional states can trigger specific physiological reactions in humans. For example, high fatigue levels are associated with an increase in blinking frequency, and elevated stress levels are reflected by a changing heart rate. Thus, physiological signals provide a means to assess and determine cognitive and emotional states of a human being. Reliable assessments of these states are crucial for enhancing performance and safety across various applications. For instance, detecting mental fatigue can improve the learning outcomes of students or prevent driving accidents in traffic. Recent advances in artificial intelligence (AI) enable better analysis of physiological signals and open up new possibilities for recognizing cognitive and emotional states. 

Literatur

• Das Chakladar, D., & Roy, P. P. (2024). Cognitive workload estimation using physiological measures: a review. Cognitive Neurodynamics, 18(4), 1445-1465
• El-Nabi, S. A., El-Shafai, W., El-Rabaie, E. S. M., Ramadan, K. F., Abd El-Samie, F. E., & Mohsen, S. (2024). Machine learning and deep learning techniques for driver fatigue and drowsiness detection: a review. Multimedia Tools and Applications, 83(3), 9441-9477.
• Mao, S., & Sejdić, E. (2022). A review of recurrent neural network-based methods in computational physiology. IEEE transactions on neural networks and learning systems, 34(10), 6983-7003.
• Rissler, R., Nadj, M., Li, M. X., Loewe, N., Knierim, M. T., & Maedche, A. (2020). To be or not to be in flow at work: physiological classification of flow using machine learning. IEEE transactions on affective computing, 14(1), 463-474.

Liste der möglichen konkreten Themen:

• AI-BA-2-1, Sommersemester 2026, Betreuung: M.Sc. Cosima von Uechtritz

  Assessing Optimal Performances in Esports Athletes

  Many gamers are familiar with the experience of becoming fully immersed in a game and losing awareness of time. This state, often referred to as the flow state, is associated with higher levels of concentration and peak performances. As such, it holds particular relevance for players seeking to enhance their performance. A deeper understanding of the flow state within professional esports could help players access it more effectively. This seminar thesis will therefore examine the characteristics and effects of the flow state in e-sports on the basis of a literature review.

  Literatur

  • Harris, D. J., Allen, K. L., Vine, S. J., & Wilson, M. R. (2023). A systematic review and meta-analysis of the relationship between flow states and performance. International review of sport and exercise psychology, 16(1), 693-721.
  • Oliveira, W., & Hamari, J. (2025). Flow Experience in Gameful Approaches: A Systematic Literature Review, Scientometric Analysis, and Research Agenda. International Journal of Human–Computer Interaction, 1-27.

• AI-BA-2-2, Sommersemester 2026, Betreuung: M.Sc. Cosima von Uechtritz

  How Can Flow Be Detected from User Interaction Data? A Systematic Literature Review

  Being highly focused and completely absorbed in a task represents a desirable state for individuals across diverse domains, including music, sport, or knowledge work. This phenomenon, also referred to as the flow state, is related to improved well-being and peak performance. To better understand these relationships, accurate and reliable measurement instruments for assessing the flow state are essential. Over the past years, a wide range of measurement approaches have emerged, such as questionnaires, physiological indicators, or behavioural data. Behavioural data include interaction logs, such as mouse movements or keystrokes, as well as performance metrics, such as reaction time, recorded during task execution. These data sources present a promising approach, as they provide a non-intrusive way to analyse flow states.

  Therefore, this seminar thesis aims to conduct a systematic literature review of existing approaches that use behavioural data to measure flow and to provide a state-of-the-art overview.

  Literatur

  • Csikszentmihalyi M., Beyond boredom and anxiety: Experiencing flow in work and play. San Francisco: Jossey-Bass, 1975.
  • Harris, D. J., Allen, K. L., Vine, S. J., & Wilson, M. R. (2023). A systematic review and meta-analysis of the relationship between flow states and performance. International review of sport and exercise psychology, 16(1), 693-721.
  • Tian, B., Zhang, S., Chen, S., Zhang, Y., Peng, K., Zhang, H., & Wang, D. (2024). Tracking dynamic flow: Decoding flow fluctuations through performance in a fine motor control task. IEEE Transactions on Affective Computing, 16(2), 891-902.
  • Wang, H. M., Hong, K. C., & Sun, C. T. (2023, November). Utilizing machine learning to predict flow state from gameplay interaction data. In 2023 IEEE Gaming, Entertainment, and Media Conference (GEM) (pp. 1-4). IEEE.

• AI-BA-2-3, Sommersemester 2026, Betreuung: M.Sc. Cosima von Uechtritz

  Performance Evaluation of Smart Health Systems

  In recent years, smart health systems have gained significant attention. Among these systems, camera-based physiological monitoring is a promising approach. Using techniques such as remote photoplethysmography (rPPG), regular RGB cameras can already estimate heart rate with high accuracy. Building on this progress, researchers now explore the potential to assess additional physiological parameters, such as blood pressure or heart rate variability. This thesis aims to provide an overview of current approaches to measuring heart rate variability using camera-based technologies, paying particular attention to their performance.

  Literatur

  • Li, J., Cheng, J., Song, R., & Liu, Y. (2025). Lst-rppg: A long-range spatio-temporal model for high-accuracy heart rate variability measurement. Expert Systems with Applications, 129526.
  • Lu, Y., Wang, C., & Meng, M. Q. H. (2020, September). Video-based contactless blood pressure estimation: A review. In 2020 IEEE international conference on real-time computing and robotics (RCAR) (pp. 62-67). IEEE.
  • Xiao, H., Liu, T., Sun, Y., Li, Y., Zhao, S., & Avolio, A. (2024). Remote photoplethysmography for heart rate measurement: A review. Biomedical Signal Processing and Control, 88, 105608.

• AI-BA-2-4, Sommersemester 2026, Betreuung: M.Sc. Cosima von Uechtritz

  The Power and Pitfalls of GenAI in Systematic Literature Reviews

  Systematic literature reviews (SLRs) are an essential part of the scientific community, providing a basis for future research, summarizing progress, and supporting theory development. Conducting an SLR involves a series of structured steps that must be followed to ensure high-quality output. Generative AI (GenAI) tools such as ChatGPT have become increasingly popular in recent years, offering promising opportunities to improve the efficiency of such structured workflows. Next to their potential, these tools still come with challenges and limitations, which can lead to misleading and erroneous outcomes. Therefore, this review aims to provide a state-of-the-art overview of existing GenAI tools, their practical applications in the context of SLRs, as well as their potential benefits and associated challenges.

  Literatur

  • Delgado-Chaves, F. M., Jennings, M. J., Atalaia, A., Wolff, J., Horvath, R., Mamdouh, Z. M., Baumbach, J., & Baumbach, L. (2025). Transforming literature screening: The emerging role of large language models in systematic reviews. Proceedings of the National Academy of Sciences, 122(2), e2411962122.
  • Galli, C., Gavrilova, A. V. & Calciolari, E. (2025). Large Language Models in Systematic Review Screening: Opportunities, Challenges, and Methodological Considerations. Information, 16(5), 378.
  • Scherbakov, D., Hubig, N., Jansari, V., Bakumenko, A. & Lenert, L. A. (2024). The emergence of Large Language Models (LLM) as a tool in literature reviews: an LLM automated systematic review. arXiv preprint arXiv:2409.04600
  • Webster, J. & Watson, R. T. (2002). Analyzing the past to prepare for the future: Writing a literature review. MIS quarterly, xiii-xxiii

Mit der zunehmenden Verfügbarkeit generativer KI entstehen neue Möglichkeiten und Herausforderungen für den Umgang mit Wissen in organisationalen und bildungsbezogenen Kontexten. Systeme wie ChatGPT verändern die Art und Weise, wie Wissen verarbeitet, aufbereitet und angewendet wird. Solche Technologien sind dazu in der Lage, komplexe kognitive Aufgaben wie Textgenerierung, Problemlösung und Informationsaufbereitung in Echtzeit zu unterstützen. Dadurch entstehen neue Formen der Interaktion mit Wissen sowie neue Möglichkeiten zur Unterstützung von Wissensaneignung und -vermittlung. So können Lehrende beispielsweise in kurzer Zeit strukturierte Lehrmaterialien oder Prüfungsaufgaben erstellen und individuelle Rückmeldungen formulieren, während Lernende generative KI beim Verfassen von Texten, beim Verständnis komplexer Inhalte oder bei der Lösung von Aufgaben nutzen. Gleichzeitig entwickelt sich die Nutzung generativer KI in vielen Bildungskontexten deutlich schneller als die entsprechenden institutionellen Regelungen, Leitlinien oder organisatorischen Rahmenbedingungen. Dadurch entsteht ein strukturelles Ungleichgewicht zwischen technologischer Entwicklung und institutioneller Steuerung, durch das informelle Nutzungspraktiken im Sinne der Schatten-KI begünstigt werden. Dieses Phänomen knüpft an das Konzept der Schatten-IT an. Während Schatten-IT die Nutzung nicht genehmigter IT-Systeme beschreibt, bezeichnet Schatten-KI die Nutzung generativer KI in Situationen, in den institutionelle Vorgaben fehlen, unklar sind oder nicht eindeutig angewendet werden. In diesen Situationen greift Schatten-KI direkt in wissensbezogene, didaktische und arbeitsbezogene Prozesse ein. Vor diesem Hintergrund stellt sich die Frage, welche Faktoren das Phänomen der Schatten-KI im Bildungswesen beeinflussen und welche Konsequenzen sich daraus ergeben.

Liste der möglichen konkreten Themen:

• APP-BA-1-1, Sommersemester 2026, Betreuung: M.Sc. Ajurthan Rameskumar

  Wo zeigt sich die Diskrepanz zwischen regelkonformer und tatsächlicher Nutzung generativer KI durch Lehrende?

  Mittlerweile setzen Lehrende zunehmend generative KI-Systeme wie ChatGPT ein, beispielsweise zur Erstellung von Lehrmaterialien und Prüfungsaufgaben oder zur Unterstützung von Rückmeldungsprozessen. Laut Studien besitzen solche Technologien das Potenzial, didaktische Prozesse effizienter zu gestalten und neue Möglichkeiten der Gestaltung von Lehrkontexten zu eröffnen (Kasneci et al., 2023; Zawacki-Richter et al, 2019). Gleichzeitig werden diese Technologien häufig in Kontexten genutzt, die noch nicht vollständig durch institutionelle Regelungen entwickelt oder klar definiert sind. Dadurch entstehen Nutzungssituationen, die nicht eindeutig durch formale Vorgaben abgedeckt sind (Christ‐Brendemühl, 2025, Fengchun & Wayne 2023). Dadurch entstehen Unsicherheiten im Umgang mit generativer KI, die wiederum die Entstehung informeller Nutzungspraktiken begünstigen. Vor diesem Hintergrund zielt die vorliegende Seminararbeit darauf ab, den aktuellen Forschungsstand zu informellen Nutzungspraktiken generativer KI im Sinne der Schatten-KI durch Lehrende im Bildungswesen durch eine systematische Literaturrecherche zu analysieren. Aufbauend darauf sollen zentrale Nutzungsmuster, Einflussfaktoren sowie zugrunde liegende Zusammenhänge identifiziert und strukturiert dargestellt werden.

  Literatur

  • Christ‐Brendemühl, S. (2025). Leveraging Generative AI in Higher Education: An Analysis of Opportunities and Challenges Addressed in University Guidelines. European Journal of Education, 60(1). doi.org/10.1111/ejed.12891
  • Fengchun, M., & Wayne, H. (2023). Guidance for generative AI in education and research. UNESCO. doi.org/10.54675/EWZM9535
  • Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103. doi.org/10.1016/j.lindif.2023.102274
  • Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on artificial intelligence applications in higher education – where are the educators? International Journal of Educational Technology in Higher Education, 16(1), 39. doi.org/10.1186/s41239-019-0171-0

• APP-BA-1-2, Sommersemester 2026, Betreuung: M.Sc. Ajurthan Rameskumar

  Wo zeigt sich die Diskrepanz zwischen regelkonformer und tatsächlicher Nutzung generativer KI durch Lernende?

  Unterstützung durch generative KI-Systeme wie ChatGPT wird bei Lernenden immer beliebter. Mögliche Anwendungsbereiche sind das Verfassen von Texten, das Verständnis komplexer Inhalte oder die Bearbeitung von Aufgaben. Laut Studien besitzen diese Technologien das Potenzial, Lerngänge effizienter zu gestalten und neue Möglichkeiten der individuellen Wissensaneignung zu eröffnen (Dwivedi et al., 2023; Kasneci et al., 2023) Gleichzeitig werden diese Technologien häufig in Kontexten eingesetzt, die noch nicht vollständig durch institutionelle Regelungen abgedeckt sind. Somit ergeben sich Nutzungssituationen, die nicht eindeutig durch formale Vorgaben abgedeckt sind (Christ‐Brendemühl, 2025, Fengchun & Wayne, 2023). Der Umgang mit generativer KI ist somit oftmals unsicher, sodass informelle Nutzungspraktiken begünstigt werden. Vor diesem Hintergrund zielt die vorliegende Seminararbeit darauf ab, den aktuellen Forschungsstand zu informellen Nutzungspraktiken generativer KI im Sinne der Schatten-KI durch Lernende im Bildungswesen mittels einer systematischen Literaturrecherche zu analysieren. Aufbauend darauf sollen zentrale Nutzungsmuster, Einflussfaktoren sowie zugrunde liegende Zusammenhänge identifiziert und strukturiert dargestellt werden.

  Literatur

  • Christ‐Brendemühl, S. (2025). Leveraging Generative AI in Higher Education: An Analysis of Opportunities and Challenges Addressed in University Guidelines. European Journal of Education, 60(1). doi.org/10.1111/ejed.12891
  • Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., Albanna, H., Albashrawi, M. A., Al-Busaidi, A. S., Balakrishnan, J., Barlette, Y., Basu, S., Bose, I., Brooks, L., Buhalis, D., … Wright, R. (2023). Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. International Journal of Information Management, 71. doi.org/10.1016/j.ijinfomgt.2023.102642
  • Fengchun, M., & Wayne, H. (2023). Guidance for generative AI in education and research. UNESCO. doi.org/10.54675/EWZM9535
  • Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103. doi.org/10.1016/j.lindif.2023.102274

• APP-BA-1-3, Sommersemester 2026, Betreuung: M.Sc. Ajurthan Rameskumar

  Welche Rolle spielen institutionelle Regelungen im Spannungsfeld zwischen regelkonformer und tatsächlicher Nutzung generativer KI im Bildungswesen?

  In Zeiten der zunehmenden Verbreitung generativer KI-Systeme wie ChatGPT sind Bildungseinrichtungen gefordert, geeignete institutionelle Regelungen, Leitlinien und organisatorische Rahmenbedingungen für den Einsatz dieser Systeme zu entwickeln (Holmes & Tuomi, 2022). Diese Regelungen sollen Orientierung im Umgang mit generativer KI schaffen, Transparenz fördern und einen verantwortungsvollen Einsatz im Bildungswesen sicherstellen. Gleichzeitig entwickeln sich institutionelle Regelungen oft langsamer als die tatsächliche Nutzung generativer KI im Bildungsalltag. Dadurch entsteht ein Spannungsfeld zwischen regelkonformer und tatsächlicher Nutzung, das durch bestehende Vorgaben nicht eindeutig aufgelöst wird (Christ‐Brendemühl, 2025, Fengchun & Wayne 2023). Die Folge ist, dass unterschiedliche Nutzungspraktiken im Sinne der Schatten-KI begünstigt werden. Vor diesem Hintergrund zielt die vorliegende Seminararbeit darauf ab, den aktuellen Forschungsstand dazu zu analysieren, welche institutionellen Regelungen und Leitlinien für generative KI im Bildungswesen existieren und wie dieser Zusammenhang mit der tatsächlichen Nutzung generativer KI sowie der Entstehung von Schatten-KI aussieht. Aufbauend darauf sollen zentrale Regelungsansätze, Einflussfaktoren sowie bestehende Forschungslücken identifiziert und strukturiert dargestellt werden.

  Literatur

  • Christ‐Brendemühl, S. (2025). Leveraging Generative AI in Higher Education: An Analysis of Opportunities and Challenges Addressed in University Guidelines. European Journal of Education, 60(1). doi.org/10.1111/ejed.12891
  • Fengchun, M., & Wayne, H. (2023). Guidance for generative AI in education and research. UNESCO. https://doi.org/10.54675/EWZM9535
  • Holmes, W., & Tuomi, I. (2022). State of the art and practice in AI in education. European Journal of Education, 57(4), 542–570. doi.org/10.1111/ejed.12533

Künstliche Intelligenz (KI) verändert nicht nur Arbeitsprozesse, sondern auch die Art und Weise, wie Menschen die eigene Rolle im Berufsleben wahrnehmen (Strich et al., 2021). KI wird immer häufiger nicht nur als Werkzeug, sondern als etwas angesehen, mit dem Menschen sich identifizieren können. Dieses Phänomen wird als KI-Identität bezeichnet und baut auf der IT-Identitätstheorie von Carter et al. (2020) auf (Qin et al., 2025). Darin wird beschrieben, wie Menschen solche Technologien als Teil der eigenen Identität verstehen können. Die KI-Identität ist ein noch junges Forschungsfeld, das wichtige Fragen für Unternehmen und Beschäftigte aufwirft. Welche Faktoren fördern oder hemmen die Entwicklung einer KI-Identität? Wie verändert sich dadurch die berufliche Rolle? Welche Chancen und Risiken ergeben sich für Organisationen? In diesem Seminarbereich setzen Sie sich mit verschiedenen Aspekten der KI-Identität auseinander. Dabei können Sie entweder eine systematische Literaturrecherche durchführen, um den Forschungsstand aufzuarbeiten, oder Interviews durchführen, um praxisnahe Einblicke zu gewinnen. Das Ziel besteht darin, ein besseres Verständnis für dieses neue Phänomen zu entwickeln und relevante Perspektiven für Wissenschaft und Praxis aufzuzeigen.

Literatur

• Carter, M., Petter, S., Grover, V. & Thatcher, J. B. (2020). Information Technology Identity: A Key Determinant of IT Feature and Exploratory Usage. MIS Quarterly, 44(3), 983–1021. doi.org/10.25300/misq/2020/14607
• Qin, M., Qiu, S., Li, S. & Jiang, Z. (2025). Research on the impact of employee AI identity on employee proactive behavior in AI workplace. Industrial Management & Data Systems. doi.org/10.1108/imds-03-2024-0211
• Strich, F., Mayer, A. & Fiedler, M. (2021). What Do I Do in a World of Artificial Intelligence? Investigating the Impact of Substitutive Decision-Making AI Systems on Employees’ Professional Role Identity. Journal Of The Association For Information Systems, 22(2), 304–324. doi.org/10.17705/1jais.00663
• Wei, S., Zhang, Y., & Dong, J. Q. (2025). Toward Artificial Intelligence Compliance:  Impacts and Mechanisms of Performance Feedback. Information Systems Research. doi.org/10.1287/isre.2023.0580

Liste der möglichen konkreten Themen:

• APP-BA-2-1, Sommersemester 2026, Betreuung: M.Sc. Dugaxhin Xhigoli

  Potenziale und Risiken der KI-Identität in Organisationen

  Viele Unternehmen setzen auf KI, um wettbewerbsfähig zu bleiben und Prozesse effizienter zu gestalten. Durch diese Entwicklung verändert sich auch die Wahrnehmung der Arbeit durch Mitarbeitende und der Stellenwert von Technologien in Bezug auf das eigene Selbstverständnis. Eine stark ausgeprägte KI-Identität kann dabei positive Impulse liefern, etwa durch mehr Engagement oder eine höhere Innovationskraft. Gleichzeitig bestehen Risiken wie Abhängigkeit, Überidentifikation oder sogar unethisches Verhalten. Für Organisationen ergibt sich daraus die zentrale Aufgabe, die Balance zwischen den Chancen und Gefahren einer starken KI-Identität zu finden. In dieser Seminararbeit werden, die mit einer starken KI-Identität verbundenen Chancen und Risiken analysiert. Mithilfe von Experteninterviews sollen sowohl positive Effekte als auch mögliche Risiken aufgezeigt werden.

  Literatur

  • Cao, L., Chen, C., Dong, X., Wang, M. & Qin, X. (2023). The dark side of AI identity: Investigating when and why AI identity entitles unethical behavior. Computers in Human Behavior, 143, 107669. doi.org/10.1016/j.chb.2023.107669
  • Richter, J. & Schaller, R. (2025). AI Identity Threats and Reinforcement in Organizations: A Theoretical Model of Professional Role Identity Implications. Annual Hawaii International Conference On System Sciences/Proceedings Of The Annual Hawaii International Conference On System Sciences. doi.org/10.24251/hicss.2025.023
  • Wei, S., Zhang, Y., & Dong, J. Q. (2025). Toward Artificial Intelligence Compliance: Impacts and Mechanisms of Performance Feedback. Information Systems Research. doi.org/10.1287/isre.2023.0580

• APP-BA-2-2, Sommersemester 2026, Betreuung: M.Sc. Dugaxhin Xhigoli

  KI-Identität zwischen Kontrollüberzeugung und Selbstwirksamkeit: Eine qualitative Untersuchung im beruflichen Kontext

  Mit der zunehmenden Verbreitung generativer KI am Arbeitsplatz verändert sich auch das berufliche Selbstverständnis der Beschäftigten. Erste empirische Hinweise deuten darauf hin, dass individuelle Kontrollüberzeugungen und soziale Einflussfaktoren die Entwicklung einer KI-Identität begünstigen. Eine solche Identität kann wiederum die wahrgenommene Selbstwirksamkeit im Umgang mit KI stärken. Allerdings bleibt bislang weitgehend unklar, wie sich diese Zusammenhänge im beruflichen Alltag konkret entfalten und welche weiteren Faktoren eine Rolle spielen. Diese Seminararbeit vertieft bestehende Erkenntnisse zur KI-Identität durch eine qualitative Interviewstudie. Mithilfe semistrukturierter Interviews mit berufstätigen KI-Nutzern und Experten soll erforscht werden, wie Beschäftigte die Beziehung zu KI erleben und welche Faktoren die Identifikation mit KI begünstigen oder behindern.

  Literatur

  • Mirbabaie, M., Brünker, F., Möllmann Frick, N. R. J., & Stieglitz, S. (2022). The rise of artificial intelligence – understanding the AI identity threat at the workplace. Electronic Markets, 32(1), 73–99. https://doi.org/10.1007/s12525-021-00496x
  • Ng, T. W. H., Sorensen, K. L., & Eby, L. T. (2006). Locus of control at work: A meta analysis. Journal of Organizational doi.org/10.1002/job.416
  • Sasongko, B., Widarni, E. L., & Bawono, S. (2020). Training Analysis and Locus of Con trol on Self Efficacy and Work Ability of Employees. HOLISTICA – Journal of Business and Public Administration, 11(1), 29–50. doi.org/10.2478/hjbpa-2020-0003
  • Wei, S., Zhang, Y., & Dong, J. Q. (2025). Toward Artificial Intelligence Compliance: Impacts and Mechanisms of Performance Feedback. Information Systems Research. doi.org/10.1287/isre.2023.0580

• APP-BA-2-3, Sommersemester 2026, Betreuung: M.Sc. Dugaxhin Xhigoli

  KI-Identität als Brücke zwischen Kompetenzerleben und innovativem Arbeitsverhalten

  Mit der flächendeckenden Integration von KI in betriebliche Arbeitsprozesse verändert sich nicht nur die Art der Arbeit, sondern auch das berufliche Selbstverständnis der Beschäftigten. Erste empirische Befunde deuten darauf hin, dass das subjektive Erleben von Kompetenz im Umgang mit KI eine zentrale Rolle dabei spielt, wie stark sich Beschäftigte mit KI als Teil der eigenen beruflichen Identität identifizieren. Zudem hängt diese Identifikation mit innovativem Arbeitsverhalten zusammen. Allerdings bleibt bislang weitgehend ungeklärt, wie sich diese psychologischen Prozesse im beruflichen Alltag konkret entfalten, welche Erfahrungen die Entwicklung einer KI-bezogenen Identität fördern oder hemmen und unter welchen organisationalen Bedingungen eine solche Identifikation tatsächlich in innovatives Handeln übergeht. Diese Seminararbeit untersucht die genannten Zusammenhänge durch eine qualitative Interviewstudie. Mithilfe semistrukturierter Interviews mit berufstätigen KI-Nutzern sowie relevanten Experten soll erforscht werden, wie Beschäftigte das Kompetenzerleben und die KI-Identifikation im Arbeitsalltag erfahren, wie diese Prozesse in innovatives Handeln münden und welche organisationalen Rahmenbedingungen dabei eine Rolle spielen.

  Literatur

  • Janssen, O. (2000). Job demands, perceptions of effort‐reward fairness and innovative work behaviour. Journal of Occupational and Organizational Psychology, 73(3), 287–302. doi.org/10.1348/096317900167038
  • Liu, B., Cheng, S., Zhou, Q., & Shi, X. (2025). The Impact of Digital Transformation Job Autonomy on Lawyers’ Support for Law Firms’ Digital Initiatives: The Mediating Role of Cognitive Adjustment and the Moderating Effect of Leaders’ Empathy. Administrative Sciences, 15(7), 260. doi.org/10.3390/admsci15070260
  • Mirbabaie, M., Brünker, F., Möllmann Frick, N. R. J., & Stieglitz, S. (2022). The rise of artificial intelligence – understanding the AI identity threat at the workplace. Electronic Markets, 32(1), 73–99. doi.org/10.1007/s12525-021-00496-x
  • Xu, J.-Q., Wu, T.-J., Duan, W.-Y., & Cui, X.-X. (2025). How the Human–Artificial Intelligence (AI) Collaboration Affects Cyberloafing: An AI Identity Perspective. Behavioral Sciences, 15(7), 859. doi.org/10.3390/bs15070859
  • Zhao, G., Luan, Y., Ding, H., & Zhou, Z. (2022). Job Control and Employee Innovative Behavior: A Moderated Mediation Model. Frontiers in Psychology, 13, 720654. doi.org/10.3389/fpsyg.2022.720654

Artificial Intelligence (AI) is increasingly recognized as a pivotal driver of transformative change across industries, reshaping enterprise processes, decision-making and value-creation (Hughes et al., 2025). Recent advances in large language models (LLMs) have enabled the emergence of a new class of AI agents, capable of reasoning, planning, and leveraging external services to execute tasks that extend beyond the limitations of traditional rule-based automation. Operating along an “continuum of agency” (Baird & Maruping, 2021, p. 318), these AI agents present both opportunities and challenges for enterprises (e.g., Stelmaszak et al., 2025; Hughes et al., 2025). This topic complex examines AI agents across four dimensions: determinants of adoption in retail enterprises, trust in AI agents in retail enterprises, risk dimensions and mitigation strategies, and explainability approaches that enhance transparency and accountability.

Literatur

• Baird, A., & Maruping, L. M. (2021). The Next Generation of Research on IS Use: A Theoretical Framework of Delegation to and from Agentic IS Artifacts. MIS Quarterly, 45(1), 315–341. https://doi.org/10.25300/misq/2021/15882
• Hughes, L., Dwivedi, Y. K., Malik, T., Shawosh, M., Albashrawi, M. A., Jeon, I., Dutot, V., Appanderanda, M., Crick, T., De, R., Fenwick, M., Gunaratnege, S. M., Jurcys, P., Kar, A. K., Kshetri, N., Li, K., Mutasa, S., Samothrakis, S., Wade, M., & Walton, P. (2025). AI Agents and Agentic Systems: A Multi-Expert analysis. Journal of Computer Information Systems, 65(4), 489–517. https://doi.org/10.1080/08874417.2025.2483832
• Stelmaszak, M., Möhlmann, M., & Sørensen, C. (2024). When Algorithms Delegate to Humans: Exploring Human-Algorithm Interaction at Uber. MIS Quarterly, 49(1), 305–330. https://doi.org/10.25300/misq/2024/17911

Liste der möglichen konkreten Themen:

• EPA-BA-1-1, Sommersemester 2026, Betreuung: Luisa Strelow , M.Sc.

  Determinants of AI Agent Adoption in Retail Enterprises

  Potential adoption scenarios of AI agents in retail enterprises encompass a broad range of intelligent, autonomous systems that support and (partially) automate decision-making processes. These include intelligent customer service assistants (e.g., chatbots; Kamoonpuri & Sengar, 2023) that handle inquiries and provide 24/7 support, inventory management agents that forecast demand and automate replenishment decisions, and dynamic pricing systems that adjust prices in real time based on demand patterns and competitor behavior. Although AI Agents promise substantial efficiency gains, cost reductions, and enhanced customer experiences, their actual adoption depends on a variety of internal and external determinants. For example, managing the interaction between humans and autonomous systems is considered as “(…) perhaps the key managerial issue of our time” (Berente et al. 2021, p. 1440). Despite the growing strategic interest in AI-driven solutions, many retail enterprises encounter significant challenges during the adoption process. Implementing AI agents requires overcoming several challenges, such as organizational change, process redesign, and the development of new capabilities. Furthermore, retailers must address potential employee resistance, ensure transparency and comply with regulatory requirements (e.g., Giroux et al., 2022). Drawing on a structured literature review, the key determinants that influence the adoption of AI agents in retail enterprises should be identified and discussed.

  Literatur

  • Berente, N., Gu, B., Recker, J., & Santhanam, R. (2021). Managing artificial Intelligence. MIS Quarterly, 45(3), 1433–1450. https://doi.org/10.25300/misq/2021/16274
  • Giroux, M., Kim, J., Lee, J. C., & Park, J. (2022b). Artificial Intelligence and Declined Guilt: Retailing morality Comparison between human and AI. Journal of Business Ethics, 178(4), 1027–1041. https://doi.org/10.1007/s10551-022-05056-7
  • Kamoonpuri, S. Z., & Sengar, A. (2023). Hi, May AI help you? An analysis of the barriers impeding the implementation and use of artificial intelligence-enabled virtual assistants in retail. Journal of Retailing and Consumer Services, 72, 103258. https://doi.org/10.1016/j.jretconser.2023.103258

• EPA-BA-1-2, Sommersemester 2026, Betreuung: Luisa Strelow , M.Sc.

  Trust in AI Agents in Retail Enterprises

  As retailers increasingly integrate AI agents into their operations, trust becomes a critical determinant of successful implementation (Foroughi et al., 2024). The mere technical deployment of AI agents does not guarantee their effective use. Rather, trust in AI agent is essential for fostering long-term acceptance, consistent usage, and reliance on their decision-making capabilities (Raut et al., 2024). Without sufficient trust, employees may ignore or override algorithmic recommendations, and customers may hesitate to engage with AI-driven interfaces, thereby limiting the potential benefits of these technologies (Huang et al., 2022). This topic therefore examines trust in AI agents from two complementary perspectives. First, the employee perspective focuses on internal users such as store managers or category managers who rely on AI agents for tasks such as pricing decision or forecasting. Their level of trust influences whether they adopt AI-generated recommendations, integrate them into their workflows, or resist them due to concerns about accuracy, fairness, or loss of control. Second, the customer perspective investigates how shoppers perceive and evaluate systems, such as chatbots, that directly influence their purchasing decisions and service experiences. In this context, trust may affect user satisfaction, engagement, and ultimately purchasing behavior. By conducting a structured literature review, the factors contributing to or undermining trust in AI agents in retail contexts should be examined, thereby establishing a foundation that enables both employees and customers to confidently and purposefully engage with AI agents.

  Literatur

  • Foroughi, B., Iranmanesh, M., Yadegaridehkordi, E., Wen, J., Ghobakhloo, M., Senali, M. G., & Annamalai, N. (2024). Factors affecting the use of ChatGPT for obtaining shopping information. International Journal of Consumer Studies, 49(1). https://doi.org/10.1111/ijcs.70008
  • Huang, R., Kim, M., & Lennon, S. (2022). Trust as a second-order construct: Investigating the relationship between consumers and virtual agents. Telematics and Informatics, 70, 101811. https://doi.org/10.1016/j.tele.2022.101811
  • Raut, G., Goel, A., & Taneja, U. (2024). Humanizing e-tail experiences: navigating user acceptance, social presence, and trust in the realm of conversational AI agents. Personal and Ubiquitous Computing, 28(6), 895–906. https://doi.org/10.1007/s00779-024-01814-8
  • Fenech, R., Baguant, P., & Ivanov, D. (2019). The changing role of human resource management in an era of digital transformation. Journal of Management Information & Decision Sciences, 22(2), 166-175.

• EPA-BA-1-3, Sommersemester 2026, Betreuung: Luisa Strelow , M.Sc.

  Risk Categories and Mitigation Strategies for AI Agents in Enterprises

  As AI agents take on an increasingly important role in decision-making, process automation, and customer interactions, enterprises need to understand and manage the risks associated with their adoption (Huang, 2025). Although AI agents offer substantial productivity gains and innovation potential, they also introduce complex organizational challenges that extend beyond traditional IT risks. They create new forms of operational, ethical, legal, and reputational exposure. Given the autonomy of AI agents, traditional frameworks for managing risk in enterprise settings may not be sufficient to address the new challenges that arise with AI agents (Leo et al., 2026). For example, biased output may lead to discriminatory decisions, while opaque model behavior can undermine accountability and transparency (Ali et al., 2025). At the same time, enterprises face evolving regulatory expectations, such as transparency, auditability, and risk management requirements (Leo et al., 2026). Without structured governance and effective mitigation strategies, enterprises risk financial loss, compliance violations, and erosion of public trust. A structured literature review therefore should be conducted to systematically identify and synthesize the key risk categories discussed in existing research and to examine the mitigation strategies proposed to address them.

  Literatur

  • Ali, M. A., Dornaika, F., & Charafeddine, J. (2025). Agentic AI: a comprehensive survey of architectures, applications, and future directions. Artificial Intelligence Review, 59(1). https://doi.org/10.1007/s10462-025-11422-4
  • Huang, K. (2025). Agentic AI. In Progress in IS. https://doi.org/10.1007/978-3-031-90026-6
  • Leo, M., Tan, F., Miao, T., & Anand, G. (2026). From threat to trust: assessing security risks of agentic AI systems. International Journal of Information Security, 25(1). https://doi.org/10.1007/s10207-025-01185-y

• EPA-BA-1-4, Sommersemester 2026, Betreuung: Luisa Strelow , M.Sc.

  Explainability Approaches for AI Agents in Enterprises

  Unlike traditional, rule-based systems, AI agents may operate as “black boxes”, making decisions or recommendations with limited visibility into their underlying reasoning (Dutta et al., 2025). As a result, users and decision-makers may struggle to understand how outputs are generated, why specific recommendations are produced, or whether the system behaves reliably and fairly. In enterprise settings, the ability to explain how AI agents make decisions is essential for ensuring that they are understood, trusted, and accepted by both employees and stakeholders (Dutta et al., 2025). Explainability enables decision-makers to verify outputs, detect errors or biases, and assess whether AI-supported decisions align with enterprise objectives (Rodriguez et al., 2025). Without adequate explainability mechanisms, AI agents may be perceived as opaque or unreliable, potentially leading to employee resistance or reduced user acceptance. This opacity poses concrete organizational challenges. Employees may be reluctant to rely on AI-supported decisions when the underlying reasoning remains unclear, thereby hindering adoption and limiting efficiency gains (Ribeiro et al., 2026). Moreover, managers and compliance officers may find it difficult to assign responsibility when decisions are (partially) delegated to autonomous systems. Regulatory developments further intensify these challenges by requiring transparency, traceability, and justification of automated decision-making processes. Explainability approaches aim to increase the interpretability of AI systems by clarifying how inputs are transformed into outputs and by identifying the factors that influence decisions. Because different stakeholders (e.g., technical experts, managers) require varying forms and levels of explanation, it is necessary to assess which approaches are appropriate for specific use cases. Therefore, a structured literature review should be conducted to identify existing explainability approaches, evaluating their strengths and limitations, and synthesizing insights on their applicability within enterprise contexts.

  Literatur

  • Dutta, P., Josan, P. K., Wong, R. K., Dunbar, B. J., Diaz-Artiles, A., & Selva, D. (2025). Are explanations helpful under uncertainty? Effects of uncertainty in AI-Assisted Spacecraft Anomaly diagnosis. Journal of Cognitive Engineering and Decision Making, 20(1), 70–95. https://doi.org/10.1177/15553434251392301
  • Ribeiro, E., Pinto, T., Reis, A., & Barroso, J. (2026). Trustworthy AI in Design: Introducing Explainable agent Systems. In Communications in computer and information science (pp. 294–304). https://doi.org/10.1007/978-3-032-15632-7_16
  • Rodriguez, S., Thangarajah, J., & Winikoff, M. (2025). Requirements-Based explainability for multi-agent systems. In Lecture notes in computer science (pp. 246–259). https://doi.org/10.1007/978-981-95-4969-6_19

Generative Artificial Intelligence (GenAI), particularly large language models (LLMs), is rapidly transforming how organizations process information, develop software, and create digital artifacts. Integrated into tools such as conversational chatbots, development environments, and automated workflows, GenAI has begun to reshape knowledge work, decision-making processes, and innovation practices. Its ability to generate human-like text, code, and design artifacts introduces new opportunities for productivity improvements, automation, and novel business models (Weber et al., 2024).

At the same time, the adoption of GenAI introduces significant socio-technical challenges (cf. Sarker et al., 2019). These include emerging cybersecurity risks such as prompt injection attacks, the potential accumulation or mitigation of technical debt through AI-assisted programming, and broader questions about the organizational business value of GenAI technologies (Chen et al., 2025; Moreschini et al., 2026). Furthermore, the integration of GenAI into the Software Development Life Cycle (SDLC) raises questions regarding productivity, quality assurance, explainability, and the evolving role of software engineers (Ruparelia, 2010).

This topic complex explores the multifaceted role of GenAI in organizations and software engineering by examining its value creation potential, its risks and limitations, and its broader socio-technical consequences. Particular attention is paid to how GenAI affects organizational information processing, cybersecurity, software development practices, and long-term system maintainability.

Literatur

• Chen, R., Feng, J., de Matos, M. G., Hsu, C., & Rao, H. R. (2025) AI-IA Nexus - Artificial Intelligence-Information Assurance Nexus: The Future of Information Systems Security, Privacy, and Quality. MIS Quarterly. misq.umn.edu/pages/call_for_papers_ai_ia
• Moreschini, S., Arvanitou, E. M., Kanidou, E. P., Nikolaidis, N., Su, R., Ampatzoglou, A., ... & Lenarduzzi, V. (2026). The Evolution of Technical Debt from DevOps to Generative AI: A multivocal literature review. Journal of Systems and Software, 231, 112599.
• Ruparelia, N. B. (2010). Software development lifecycle models. ACM SIGSOFT Software Engineering Notes, 35(3), 8-13.
• Sarker, S., Chatterjee, S., Xiao, X., & Elbanna, A. (2019). The sociotechnical axis of cohesion for the IS discipline: its historical legacy and its continued Relevance. MIS quarterly, 43(3), 695-719.
• Weber, T., Brandmaier, M., Schmidt, A., and Mayer, S.. 2024. Significant Productivity Gains through Programming with Large Language Models. Proc. ACM Hum.-Comput. Interact. 8, EICS, https://doi.org/10.1145/3661145

Liste der möglichen konkreten Themen:

• EPA-BA-2-1, Sommersemester 2026, Betreuung: Michael Dominic Harr , M.Sc.

  Mapping the Threat Landscape of Generative AI in Organizations: A Literature Review

  Generative artificial intelligence (GenAI), in particular large language models (LLMs) embedded in chatbots, software development environments, or agentic workflows, is rapidly becoming part of everyday organizational information processing. Arguably, GenAI has emerged as a general-purpose technology that fundamentally alters how work is coordinated, knowledge is created/curated, and innovation unfolds across industries (Storey et al., 2025). Yet the same characteristics that make GenAI valuable in organizations – interaction in (human) natural language, broad generalization across tasks, and easy integration into enterprise systems, tools (e.g., Copilot), and data sources – also expand the cyber-attack possibilities in ways that are currently not well explored. An important distinction stemming from GenAI is that attacks can be linguistic and contextual, rather than purely syntactic. For instance, direct prompt injection attacks – providing instructions that ignore LLMs intended instructions by the organization or LLM provider to achieve desired goals – emerged as a salient opportunity for malicious attempts (e.g., for retrieving system level prompts or hidden instructions from the organization). Indirect prompt injections can occur when malicious instructions are planted in sources the model later retrieves (e.g., web pages, internal documents, emails, etc.), enabling hijacking without a direct interface to the organizational GenAI system (Naik et al., 2025).

  Information systems research has just begun to document such threats, mechanisms, and outcomes, where manipulated inputs could degrade integrity and decision quality (e.g., Godasu & Young, 2025). Despite growing recognition of this topic as demonstrated by recent calls for papers (e.g., Chen et al., 2025), a synthesis of cybersecurity risks emerging from GenAI in organizations is missing. Accordingly, the aim of the seminar paper is to identify, synthesize, and classify cybersecurity risks/threats that emerge from GenAI in organizations by means of a literature review (it might be useful to integrate gray literature as well).

  Literatur

  • Chen, R., Feng, J., de Matos, M. G., Hsu, C., & Rao, H. R. (2025) AI-IA Nexus - Artificial Intelligence-Information Assurance Nexus: The Future of Information Systems Security, Privacy, and Quality. MIS Quarterly. misq.umn.edu/pages/call_for_papers_ai_ia
  • Godasu, Rajesh and Young, Jacob, "Textual and Visual Prompt Injection Attacks on Large Vision-Language Models in Medical Imaging" (2025). International Conference on Information Systems (ICIS) 2025 TREOS. 172. aisel.aisnet.org/treos_icis2025/172
  • Naik, D., Naik, I., & Naik, N. (2025). When Generative AI Prompts Bite Back: Investigating Different Types of Prompt Injection Attacks on Large Language Models (LLMs) and Their Prevention Methods. Authorea Preprints.
  • Storey, V. C., Yue, W. T., Zhao, J. L., & Lukyanenko, R. (2025). Generative artificial intelligence: Evolving technology, growing societal impact, and opportunities for information systems research. Information Systems Frontiers, 1-22.

• EPA-BA-2-2, Sommersemester 2026, Betreuung: Michael Dominic Harr , M.Sc.

  Friend or Foe? A State-of-the-Art Review on Generative Artificial Intelligence and Technical Debt

  In software engineering, technical debt may be understood as the implied future cost of rework, maintenance, and reduced development speed due to choosing quick, easy, temporary, or suboptimal solutions over better, more robust code (Cunningham, 1992). Information Systems (IS) research demonstrated that technical debt is not only a technical phenomenon but also a socio-technical trap: organizations can become trapped between short-term pressures and long-term system viability, making technical debt difficult to unwind once it is embedded in practices and architectures (Rinta-Kahila et al., 2023). With the uprise of generative artificial intelligence (GenAI), organizations now increasingly face a GenAI-induced paradox (e.g., Moreschini et al., 2026). On the one hand, GenAI may act as a contributor to new or amplified technical debt (e.g., quality, maintainability, workflow side-effects) with software developers “vibe coding” so called “AI slop” (i.e., “digital content of low quality that is produced usually in quantity by means of artificial intelligence”) – the 2025 word of the year according Feldman (2025). On the other hand, research has shown that GenAI may be employed to detect and classify technical debt, repay or refactor certain code snippets, or to increase test coverage and assist in behavioral test generation (i.e., debt-reduction) in order to reduce overall technical debt. Detection and classification of technical debt is, however, heavily focused on self-admitted technical debt (SATD; e.g., when developers annotate/comment code snippets with “TODO” or “quick workaround, fix later”). While research is still fragmented and in heavy discussions, the net impact of GenAI on technical debt is plausibly bi-directional: GenAI can reduce debt by accelerating remediation, but it can also increase debt by scaling production of code and artifacts that later require interpretation and cleanup. Hence, the aim of this seminar is to map the state of the art on GenAI and technical debt based on a systematic literature review (e.g., contexts, units of analysis, outcomes), classify the literature into a structured framework (e.g., GenAI for detection/classification, GenAI for remediation, GenAI induced technical debt, Technical debt in GenAI) while explicitly differentiating between code-, architecture-, and GenAI-levels. The seminar should close with an IS-relevant research agenda.

  Literatur

  • Cunningham, W. (1992). The WyCash portfolio management system. ACM Sigplan Oops Messenger, 4(2), 29-30.
  • Feldman, E. (2025). Merriam-Webster’s Word of the Year for 2025 Is ‘Slop,’ the A.I.-Generated Junk That Fills Our Social Media Feeds. Smithsonian Magazine.
  • Moreschini, S., Arvanitou, E. M., Kanidou, E. P., Nikolaidis, N., Su, R., Ampatzoglou, A., ... & Lenarduzzi, V. (2026). The Evolution of Technical Debt from DevOps to Generative AI: A multivocal literature review. Journal of Systems and Software, 231, 112599.
  • Rinta-Kahila, T., Penttinen, E., & Lyytinen, K. (2023). Getting trapped in technical debt: Sociotechnical analysis of a legacy system’s replacement. Mis Quarterly, 47(1), 1-32.

• EPA-BA-2-3, Sommersemester 2026, Betreuung: Frederik Hendricks , M.Sc.

  Identifying the Business Value of Generative Artificial Intelligence Technologies

  Artificial Intelligence (AI) is a broad field of computer science concerned with creating systems that can perform tasks that traditionally required human intelligence, such as perception, reasoning, learning, and problem-solving (Russell, 2016). Generative Artificial Intelligence (GenAI) is a novel and rapidly evolving branch of AI that focuses on creating new content such as text, images, audio, or software code, based on patterns learned from large-scale datasets (Feuerriegel et al., 2023). Unlike traditional AI systems, which primarily classify or predict, GenAI systems can produce novel outputs that often resemble human-created artifacts (Feuerriegel et al., 2023). This creative capacity has attracted significant attention in both academic and business contexts.

  Potential application areas are broad and expanding. Use cases can range between the automatic generation of product descriptions (Ghaffari et al., 2024) or building new business models (Kanbach et al., 2024). It shows that GenAI can be used to speed up processes and to allow for the persuasion of new business models. 

  However, the IT business value (Schryen, 2013) of GenAI, i.e., the benefit an organization gains from IT, remains subject to debate. The effects of GenAI on organizational performance have not yet been systematically categorized. This lack of structured understanding makes it difficult to distinguish which impacts are specific to GenAI and which overlap with other IT-based innovations. The topic, therefore, aims to identify, categorize, and critically reflect on the impacts of GenAI to provide a clearer overview and better differentiation.

  Literatur

  • Feuerriegel, S., Hartmann, J., Janiesch, C., & Zschech, P. (2023). Generative AI. Business & Information Systems Engineering, 66(1), 111–126. doi.org/10.1007/s12599-023-00834-7
  • Ghaffari, S., Yousefimehr, B., & Ghatee, M. (2024). Generative-AI in E-Commerce: Use-Cases and Implementations. 2024 20th CSI International Symposium on Artificial Intelligence and Signal Processing (AISP), 1–5. doi.org/10.1109/aisp61396.2024.10475266
  • Kanbach, D. K., Heiduk, L., Blueher, G., Schreiter, M., & Lahmann, A. (2023). The GenAI is out of the bottle: generative artificial intelligence from a business model innovation perspective. Review of Managerial Science, 18(4), 1189–1220. doi.org/10.1007/s11846-023-00696-z
  • Russell, S. (2016). Artificial intelligence (P. Norvig, Ed.; Third edition.). Pearson.
  • Schryen, G. (2013). Revisiting IS business value research: what we already know, what we still need to know, and how we can get there. European Journal of Information Systems, 22(2), 139–169. https://doi.org/10.1057/ejis.2012.45

• EPA-BA-2-4, Sommersemester 2026, Betreuung: Pierre Maier , M.Sc.

  LLM-Assisted Model-Driven Software Development: Prospects and Remaining Challenges

  Model-driven software development is an approach to software development where models are treated as first-class citizens. Developers primarily interact with conceptual models (typically in the form of diagrams) and only add/modify the code directly where required. Models are transformed into executable programming code. Model-driven software development is also intended to empower end users in supporting the development of the software they are using. The more recent ‘low-code’/’no-code’ movements can be considered approaches to model-driven software development (Cabot 2022).

  LLMs are appealing, among others, because they offer users a natural-language interface with a seemingly open-ended set of queries (cf. Petroni et al. 2019). One of the many prospects of LLMs is the (partial) automation of software-development activities such as development and testing. This opens up new perspectives for model-driven software development. 

  This seminar thesis has two main objectives. First, you should develop/adopt a framework of model-driven software development that expresses the primary aims and requirements for each step in the development process. Second, you should analyze/synthesize how the use of LLMs is recommended for which step and what challenges/limitations they encounter. You should pay special attention to (necessary/sufficient) conditions that must be met for the adequate use of LLMs.

  Literatur

  • Almonte L, Guerra E, Cantador I, de Lara J (2022) Recommender Systems in Model-Driven Engineering: A Systematic Mapping Review. Software and Systems Modeling 21:249–280
  • Cabot J (2020) Positioning of the Low-Code Movement within the Field of Model-Driven Engineering. Proceedings of the 23rd ACM/IEEE International Conference on Model-Driven Engineering Languages and Systems: Companion Proceedings
  • da Silva, Alberto Rodrigues (2015) Model-Driven Engineering: A Survey Supported by the Unified Conceptual Model. Computer Languages, Systems & Structures 43:139–155
  • Dalianis H (1992) A Method for Validating a Conceptual Model by Natural Language Discourse Generation. In: Loucopolous P (ed). Advanced Information Systems Engineering, CAiSE 1992, vol 141. Springer International Publishing: Cham, pp 425–444
  • Fill H-G, Fettke P, Köpke J (2023) Conceptual Modeling and Large Language Models: Impressions From First Experiments With ChatGPT. Enterprise Modelling and Information Systems Architectures 18(3):1–15
  • Petroni F, Rocktäschel T, Lewis P, Bakhtin A, Wu Y, Miller AH, Riedel S (2019) Language Models as Knowledge Bases? Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pp 2463–2473
  • See also the references listed for the seminar thesis “The Role of Generative AI in Software Development”

• EPA-BA-2-5, Sommersemester 2026, Betreuung: Frederik Hendricks , M.Sc.

  The role of Generative Artificial Intelligence in Software Development

  Generative Artificial Intelligence (GenAI) has emerged as a transformative force in software engineering. With the advent of large language models such as ChatGPT and AI-powered development tools like GitHub Copilot, software development processes are increasingly augmented by machine-generated content (Sobania et al., 2023; Song et al., 2024). While much attention has been given to AI-supported code generation, its broader role across the Software Development Life Cycle (SDLC) remains insufficiently systematized in academic literature.

  The modalities and influences of GenAI application across the SDLC, as well as the criteria by which its prospective contributions to key software engineering activities might be assessed, remain insufficiently understood. Particular emphasis should be placed on early-phase activities such as the generation of user stories, derivation and refinement of requirements, formalization of use cases, and the automated creation of design artifacts (e.g., ER diagrams and architectural models). In addition, GenAI’s role in code generation, refactoring, automated test creation, documentation, and knowledge management should be analyzed.

  The aim of this thesis is to conduct a structured literature review and critically assess both the opportunities and limitations of GenAI in regard to its software development capabilities. Relevant dimensions include but are not limited to productivity gains, quality implications, explainability, hallucination risks, traceability of requirements, data privacy concerns, and the evolving role of software engineers. Empirical findings on AI-assisted programming productivity (e.g., Noy & Zhang, 2023) and emerging discussions on AI in requirements engineering and design automation should be integrated.

  Literatur

  • Noy, S., & Zhang, W. (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science.
  • Ruparelia, N. B. (2010). Software development lifecycle models. ACM SIGSOFT Software Engineering Notes, 35(3), 8-13.
  • Santos, P. D. O., Figueiredo, A. C., Nuno Moura, P., Diirr, B., Alvim, A. C., & Santos, R. P. D. (2024, May). Impacts of the usage of generative artificial intelligence on software development process. In Proceedings of the 20th Brazilian Symposium on Information Systems (pp. 1-9).
  • Sobania, D., Briesch, M., Hanna, C. and Petke J. (2023) An Analysis of the Automatic Bug Fixing Performance of ChatGPT, 2023 IEEE/ACM International Workshop on Automated Program Repair (APR), pp. 23-30, doi: 10.1109/APR59189.2023.00012.
  • Song, F., Agarwal, A., & Wen, W. (2024). The impact of generative AI on collaborative open-source software development: Evidence from GitHub Copilot. arXiv preprint arXiv:2410.02091.

• EPA-BA-2-6, Sommersemester 2026, Betreuung: Clemens Brackmann , M.Sc.

  Generative Artificial Intelligence in critical decision making - How can Generative AI support dynamic pricing decisions in stationary retail?

  Pricing is one of the most strategically critical decisions in retail management. Traditional approaches such as cost-plus or competitive pricing models are inherently static and fail to respond to real-time demand fluctuations, competitive moves, or shifting consumer behavior (Den Boer, 2015). While AI-based dynamic pricing has emerged as a promising solution, its adoption in stationary retail significantly lags behind e-commerce, where algorithmic pricing is already well established (Assad et al., 2024). Stationary retailers face unique challenges: they operate across physical store environments, manage large and heterogeneous product assortments, and must balance pricing efficiency with customer-facing transparency.

  Generative AI introduces a qualitatively new dimension to this problem. Unlike classical machine learning models that require large labeled datasets and technical expertise, GenAI enables pricing support through natural language interaction, substantially lowering adoption barriers for practitioners (Basal & Saraç, 2024; Cohen, 2026). This raises an important and underexplored question: can GenAI serve as a viable decision support tool for pricing managers in stationary retail — and if so, under what conditions? The existing literature on AI-driven pricing largely focuses on technical optimization in digital retail environments and has not yet systematically examined the role of GenAI as a managerial decision support tool in physical retail contexts (Spann et al., 2025). This gap motivates a structured literature review that synthesizes existing knowledge, contrasts GenAI with traditional ML-based approaches, and derives an agenda for future research in this practically relevant domain.

  Literatur

  • den Boer, A. V. (2015). Dynamic pricing and learning: Historical origins, current research, and new directions. Surveys in Operations Research and Management Science, 20(1), 1–18. https://doi.org/10.1016/j.sorms.2015.03.001
  • Basal, M., & Saraç, E. (2024). Dynamic pricing strategies using artificial intelligence algorithms. Open Journal of Business and Management, 12, 1–18. Scientific Research Publishing. doi.org10.4236/ojapps.2024.148128
  • Cohen, M. C. (2026). How to use generative AI for pricing. MIT Sloan Management Review, January 2026. https://sloanreview.mit.edu/article/how-to-use-generative-ai-for-pricing/
  • Spann, M., Bertini, M., Koenigsberg, O., Zeithammer, R., Aparicio, D., Chen, Y., Fantini, F., Jin, G. Z., Morwitz, V. G., Popkowski Leszczyc, P., Vitorino, M. A., Yalcin Williams, G., & Yoo, H. (2025). Algorithmic pricing: Implications for marketing strategy and regulation. International Journal of Research in Marketing.https://doi.org/10.1016/j.ijresmar.2025.05.001
  • Assad, S., Clark, R., Ershov, D., & Xu, L. (2024). Algorithmic pricing and competition: Empirical evidence from the German retail gasoline market. Journal of Political Economy, 132(3), 723–771. https://doi.org/10.1086/726906

• EPA-BA-2-7, Sommersemester 2026, Betreuung: Clemens Brackmann , M.Sc.

  Human-AI collaboration in retail pricing decisions: A literature review on trust, control, and accountability

  As AI systems increasingly penetrate retail pricing processes, a critical organizational and managerial challenge emerges: how should human decision-makers interact with, oversee, and take responsibility for AI-generated pricing recommendations? This question is particularly pressing in stationary retail, where pricing decisions directly affect customer relationships, competitive positioning, and brand perception — consequences that extend far beyond a misplaced algorithm.

  The risks of unchecked automation are well documented. Real-world failures such as an Amazon book being algorithmically priced at over $23 million illustrate what happens when autonomous pricing systems operate without meaningful human oversight (Basal & Saraç, 2024). At the same time, purely human pricing decisions are slow, inconsistent, and unable to process the volume of data modern retail demands (Wu et al., 2022). This creates a fundamental tension: organizations need AI to scale and optimize their pricing, yet full delegation to autonomous systems introduces accountability gaps and fairness concerns that neither managers nor customers are willing to accept (Bar-Gill et al., 2023).

  Research confirms that human involvement in AI-assisted decisions significantly improves perceived fairness and trust compared to fully automated outcomes (Schoeffer et al., 2022). However, excessive human reliance on AI recommendations introduces the risk of automation bias — the uncritical acceptance of system outputs regardless of their quality (Lee & See, 2004; Dietvorst et al., 2015). Despite its practical urgency, this tension between efficiency and human control in retail pricing remains theoretically underdeveloped, particularly for stationary retail contexts. This paper addresses the gap by synthesizing the literature on trust, control, and accountability in human-AI pricing systems.

  Literatur

  • Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors: The Journal of the Human Factors and Ergonomics Society, 46(1), 50–80. https://doi.org/10.1518/hfes.46.1.50_30392
  • Schoeffer, J., Kuehl, N., & Machowski, Y. (2022). "There is not enough information": On the effects of explanations on perceptions of informational fairness and trustworthiness in automated decision-making. In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency (FAccT '22) (pp. 1–14). Seoul, Republic of Korea. ACM. https://doi.org/10.1145/3531146.3533202
  • Schoeffer, J., De-Arteaga, M., & Kühl, N. (2024). Explanations, fairness, and appropriate reliance in human-AI decision-making. In Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI '24). Honolulu, HI, USA. ACM. https://doi.org/10.1145/3613904.3642621
  • Dietvorst, B. J., Simmons, J. P., & Massey, C. (2015). Algorithm aversion: People erroneously avoid algorithms after seeing them err. Journal of Experimental Psychology: General, 144(1), 114–126. https://doi.org/10.1037/xge0000033
  • Wu, Z., Yang, Y., Zhao, J., & Wu, Y. (2022). The impact of algorithmic price discrimination on consumers' perceived betrayal. Frontiers in Psychology, 13, 825420. https://doi.org/10.3389/fpsyg.2022.825420
  • Bar-Gill, O., Sunstein, C. R., & Talgam-Cohen, I. (2023). Algorithmic harm in consumer markets. Journal of Legal Analysis, 15(1), 1–47. https://doi.org/10.1093/jla/laad003

Generative Artificial Intelligence (GenAI), particularly large language models (LLMs), has rapidly become a go to foundation – maybe even an imperative – of digital tools that students draw on to accomplish academic tasks. Recent reports found that four out of five students are using GenAI during their studies (QS, 2023), mostly to explain concepts, summarize articles, and suggest research ideas (Freeman, 2025). In this context, universities can be understood as socio-technical contexts in which learning and performance emerge from configurations of tasks and technologies (i.e., GenAI) that are inevitably embedded in institutional arrangements (e.g., course designs, assessments, integrity norms; see Sarker et al., 2019). This topic complex therefore examines GenAI not only as a “novel tool,” but as an inevitably intertwined socio-technical information system that reshapes how students’ work is produced, validated, and accounted for, with direct implications for Information Systems curricula and competence development (Van Slyke et al., 2023). Importantly, however, GenAI is not conceived as an “automatic learning enhancement”, because effective use may require prompting, verifying, and revising, which imposes metacognitive demands on students that can affect learning and create risks of overreliance (Tankelevitch et al., 2024). “Effective use” requires responsible use, which is not only an individual capability but also an outcome of above-mentioned institutional arrangements. Following this, GenAI constitutes a major challenge for German Universities: making GenAI accessible for research, teaching, and learning in ways that are legally compliant, economically sustainable, but also compatible with educational goals, are major concerns. The University of Duisburg–Essen (UDE), for example, exemplifies an integration-oriented response to these sector-wide challenges by providing university members with institutionally mediated access to “Chat AI” via its KI-Portal, complemented by locally defined guidance and governance constraints (UDE, 2026). Yet, empirical evidence remains limited regarding how students assemble and use GenAI tool repertoires, how they align specific tools with academic goals, and to what extent they are sensitized to principles of responsible use in their study practices. Accordingly, this topic complex addresses these gaps to develop an empirically grounded foundation for curriculum development, targeted support measures, and integrity-preserving approaches to IS education.

Literatur

• Freeman, J. (2025). Student Generative AI Survey 2025 [Report]. Accessed March 5^th, 2026. Retrieved from: www.hepi.ac.uk/wp-content/uploads/2025/02/HEPI-Kortext-Student-Generative-AI-Survey-2025.pdf.
• UDE (2026). KI-Portal – Chat AI für Hochschulangehörige [Online]. Accessed March 5^th, 2026. Retrieved from: https://www.uni-due.de/de/digitalisierung/ki-portal/chat-ai-login.php.
• QS (2023). Universities, students and the Generative AI Impact [Report]. Accessed March 5^th, 2026. Retrieved from: https://tinyurl.com/2uz9h87m.
• Sarker, S., Chatterjee, S., Xiao, X., & Elbanna, A. (2019). The sociotechnical axis of cohesion for the IS discipline: its historical legacy and its continued Relevance1. MIS quarterly, 43(3), 695-719.
• Tankelevitch, L., Kewenig, V., Simkute, A., Scott, A. E., Sarkar, A., Sellen, A., & Rintel, S. (2024, May). The metacognitive demands and opportunities of generative AI. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (pp. 1-24).
• Van Slyke, C., Johnson, R. D., & Sarabadani, J. (2023). Generative artificial intelligence in information systems education: Challenges, consequences, and responses. Communications of the Association for Information Systems, 53(1), 1-21.

Liste der möglichen konkreten Themen:

• EPA-BA-3-1, Sommersemester 2026, Betreuung: Michael Dominic Harr , M.Sc.

  GenAI Usage by Information Systems Students: An Interview Study

  Generative AI (GenAI), in particular large language models (LLMs) embedded in tools such as chatbots, writing assistants, and coding copilots, has become a routine part of students’ day-to-day study practices. In our exercises for lectures, seminar papers, as well as theses, we see GenAI usage increasing in general. In Information Systems (IS) education, researchers argue that GenAI will affect not only what students learn (e.g., analysis, modeling, development) but also how they learn and demonstrate competence, creating new challenges around academic integrity, skill development, and pedagogical design (Van Slyke et al., 2023). However, what is still insufficiently understood, especially at the level useful for curriculum design and policy, is how IS students actually use GenAI: which tools they rely on, how they match tools to goals (e.g., ideation vs. explanation vs. coding), and what strategies they use to judge output quality (see for instance Sun & Deng, 2025). This is important to know, because recent research suggests that GenAI use does not “automatically help” students in their understanding; it imposes substantial metacognitive demands (e.g., planning prompts, monitoring output, verifying, revising), which can shape learning quality and maybe overreliance (Tankelevitch et al., 2024).

  This seminar addresses that issue with semi-structured interviews (e.g., Adeoye‐Olatunde & Olenik, 2021) focusing on IS students’ tool repertoires (which tools they use), goal–tool alignment (for which problem spaces / goals they use such tools), and reflection (how they make sense of the outputs and judge them). It would be worthwhile to then map the goal-tool alignments with opportunities and challenges. Hence, the aim is to develop a grounded understanding of how IS students use which GenAI tools, for which goals, and how they interact with them and judge the output generated.

  Literatur

  • Adeoye‐Olatunde, O. A., & Olenik, N. L. (2021). Research and scholarly methods: Semi‐structured interviews. Journal of the american college of clinical pharmacy, 4(10), 1358-1367.
  • Sun, R., & Deng, X. (2025). Using generative AI to enhance experiential learning: An exploratory study of ChatGPT use by university students. Journal of Information Systems Education, 36(1), 53-64.
  • Tankelevitch, L., Kewenig, V., Simkute, A., Scott, A. E., Sarkar, A., Sellen, A., & Rintel, S. (2024, May). The metacognitive demands and opportunities of generative AI. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (pp. 1-24).
  • Van Slyke, C., Johnson, R. D., & Sarabadani, J. (2023). Generative artificial intelligence in information systems education: Challenges, consequences, and responses. Communications of the Association for Information Systems, 53(1), 1-21.

• EPA-BA-3-2, Sommersemester 2026, Betreuung: Michael Dominic Harr , M.Sc.

  Sensibilisierung von Bachelorstudierenden beim Einsatz generativer künstlicher Intelligenz im Studium: Eine Interviewstudie

  Generative künstliche Intelligenz (GenAI), insbesondere Large Language Models (LLMs), die in Chatbots, Schreib- und Codingtools verfügbar sind (z. B. Copilot, ChatGPT, NotebookLM), werden zunehmend nicht nur von Unternehmen sondern insbesondere von Schülern und Studierenden eingesetzt. GenAI hat sich dabei in den letzten Jahren zu einem routinierten Partner im täglichen universitären Leben etabliert (vgl. Park, 2025). In unseren Modulen, wie beispielsweise Enterprise Transformation, aber auch bei Seminararbeiten, Projektarbeiten und Abschlussarbeiten nehmen wir immer stärker wahr, dass Studierende für Ihr Studium GenAI einsetzen. Entsprechend rückt in der Wirtschaftsinformatiklehre weniger die Frage in den Vordergrund, ob Studierende GenAI nutzen, sondern wie sie diese Systeme in Lernprozesse integrieren, welche Fähigkeiten dafür erforderlich sind und inwiefern die Studierenden hinsichtlich des Einsatzes von GenAI sensibilisiert sind (Van Slyke et al., 2023). GenAI zeichnet sich durch eine Vielzahl an Fähigkeiten aus (vgl. Harr et al., 2024), die im Studium Chancen (z. B. individuelle/personalisierte Unterstützung, schnellere Iterationen) aber auch Risiken (z. B. plausible, aber falsche Inhalte, Abhängigkeit, Integritäts- und Qualitätsprobleme) zugleich verstärken können; was sich wiederum in neuen didaktischen Herausforderungen ausdrückt (siehe Gimpel et al., 2025). Die Nutzung von GenAI ist nicht per se problematisch und an der Universität Duisburg-Essen prinzipiell erlaubt, kann aber bei fehlender Sensibilisierung zu problematischen Fehlanwendungen führen; etwa durch ungeprüfte Übernahme halluzinierter Inhalte.

  Ziel des Seminars ist deshalb – basierend auf semi-strukturierten Interviews (vgl. Übersicht von Iyamu, 2018) – den Status quo der Sensibilisierung von Bachelorstudierenden der Wirtschaftsinformatikbeim Einsatz von GenAI im Studium systematisch zu erfassen. Basierend auf reichhaltigen Erkenntnissen der Interviews können so Implikationen für die Hochschullehre und Unterstützungsangebote abgeleitet werden (z. B. Schulungen, Trainings, andere Aufgaben- und Klausurformate, etc.).

  Literatur

  • Gimpel, H., Hall, K., Decker, S., Eymann, T., Gutheil, N., Lämmermann, L., Braig, N., Maedche, A., Röglinger, M., Ruiner, C., Manfred Schoch, Schoop, M., Urbach, N., & Vandirk, S. (2025). Using Generative AI in Higher Education: A Guide for Instructors. Journal of Information Systems Education, 36(3), 237-256. https://doi.org/10.62273/QLLG7172
  • Harr, M. D., Wienand, M., & Schütte, R. (2024). Towards Enhanced E-Learning Within MOOCs: Exploring the Capabilities of Generative Artificial Intelligence. In Proceedings of the Pacific-Asia Conference on Information Systems.
  • Iyamu, T. (2018). Collecting qualitative data for information systems studies: The reality in practice. Education and Information Technologies, 23(5), 2249-2264.
  • Park, J. (2025). A systematic literature review of generative artificial intelligence (GenAI) literacy in schools. Computers and Education: Artificial Intelligence, 100487.
  • Van Slyke, C., Johnson, R. D., & Sarabadani, J. (2023). Generative artificial intelligence in information systems education: Challenges, consequences, and responses. Communications of the Association for Information Systems, 53(1), 1-21.

Organizations increasingly rely on data-driven technologies to improve operational efficiency, gain deeper insights into business processes, and support strategic decision-making. Emerging technologies such as the Internet of Things (IoT) and Artificial Intelligence (AI) are expanding the ways in which data can be collected, analyzed, and used to create organizational value (Lv et al., 2021; Shankar, 2023).

IoT technologies enable the continuous collection of real-time data from physical environments through sensors, RFID systems, and connected devices. In domains such as retail, these technologies provide detailed insights into inventory levels, product movements, and customer behavior, allowing organizations to optimize operations and improve supply chain flexibility (Hossain et al., 2021). At the same time, advances in AI, particularly generative AI and natural language interfaces, are transforming how users interact with organizational data. Instead of relying on static dashboards and predefined analytics, AI-enabled systems allow users to explore data interactively through natural language queries and dialogue-based interfaces (Awad et al., 2025).

This topic complex explores how digital technologies reshape organizational data practices and decision support systems. It focuses on the economic impact, technological capabilities, and organizational challenges associated with leveraging IoT and AI for data-driven value creation.

Literatur

• Lv, Z., Qiao, L., Verma, S., & Kavita. (2021). AI-enabled IoT-edge data analytics for connected living. ACM Transactions on Internet Technology, 21(4), 1-20.
• Shankar, A. (2023). Efficient data interpretation and artificial intelligence enabled IoT based smart sensing system. Artificial Intelligence Review, 56(12), 15053-15077.
• Hossain, M. S., Chisty, N. M. A., Hargrove, D. L., & Amin, R. (2021). Role of Internet of Things (IoT) in retail business and enabling smart retailing experiences. Asian Business Review, 11(2), 75-80.
• Awad, M. N. A., Ivanov, S., Tikhonova, O., & Khodnenko, I. (2025). A Multimodal Conversational Agent for Tabular Data Analysis. arXiv preprint arXiv:2511.18405.

Liste der möglichen konkreten Themen:

• EPA-BA-4-1, Sommersemester 2026, Betreuung: Frederik Hendricks , M.Sc.

  Profitability of Internet of Things (IoT) Applications in Retail: Opportunities, Economic Impact, and Conditions for Value Creation

  The Internet of Things (IoT) is increasingly transforming retail environments through technologies such as RFID tracking, smart shelves, sensors, and automated monitoring systems. These technologies enable real-time data collection on inventory levels, product movement, and customer behavior, thereby improving operational efficiency and decision-making in retail stores. For example, IoT-based smart shelf systems can automatically detect product availability and trigger restocking processes, which significantly improves inventory management and reduces human error (AlQahatani et al., 2025).

  Beyond operational improvements, IoT technologies also enable new analytical capabilities by integrating sensor data with transactional information. This allows retailers to analyze customer journeys, optimize product placement, and improve supply chain visibility, potentially increasing sales performance and customer satisfaction.

  However, the implementation of IoT solutions in retail is associated with substantial investments in hardware, software integration, and data management infrastructure. Furthermore, challenges such as security risks, system scalability, vendor lock-in effects, and device heterogeneity can influence the economic viability of these technologies (Roe et al., 2022).

  The aim of this thesis is to analyze the economic impact and potentials of IoT in the retail domain under consideration of the necessary conditions that must be given, before actual value can be created by these technologies.

  Literatur

  • AlQahtani, A.A.S., Darrat, A.A., Turpin, L. et al. Smart shelves: transforming retail stocking with internet of things and machine learning. J. Umm Al-Qura Univ. Eng.Archit. 16, 1864–1880 (2025). doi.org/10.1007/s43995-025-00213-1
  • Feng, Z., & Zhang, Z. (2024). IoT in retail: Transforming big data analytics for business success. International Journal of Engineering and Science Invention (IJESI), Vol13, (9), 69-75.
  • Roe, M., Spanaki, K., Ioannou, A., Zamani, E. D., & Giannakis, M. (2022). Drivers and challenges of internet of things diffusion in smart stores: A field exploration. Technological Forecasting and Social Change, 178, 121593.
  • Sherovska, G. (2023). Internet of Things (IoT) Applications in the Retail Sector: A Focus on the FMCG Industry. Journal of Economic Development, Environment and People, 12(1), 16-29.
  • Team, M. I. (2021). How IoT tracking in retail supply chains drives ROI. https://www.microsoft.com/en-us/industry/blog/retail/2021/04/01/how-iot-tracking-in-retail-supply-chains-drives-roi/ abgerufen am 06.03.2025

• EPA-BA-4-2, Sommersemester 2026, Betreuung: Frederik Hendricks , M.Sc.

  Interactive Data Exploration: How AI Transforms Data Exploration Beyond Static Dashboards

  Traditional business intelligence systems rely heavily on static or semi-interactive dashboards that require users to navigate predefined visualizations and metrics (Awad et al., 2025). They are often focused on windows, icons, menus, and pointer interaction (Peng et al., 2022). While these dashboards have long been a central tool for data analysis and decision support, their limitations become increasingly evident in dynamic, complex, and data-rich environments. Recent advances in generative artificial intelligence (GenAI) open up new possibilities for data interaction by enabling natural-language, dialogue-oriented exploration of data sets. Instead of adapting questions to fixed dashboards, users can interact with data in a more flexible, intuitive, and context-aware manner (Awad et al., 2025). 

  The aim of this seminar paper is to evaluate of how data analytics systems that leverages generative AI can support exploratory data analysis and partially or completely replace classic dashboard interactions. A central focus of the thesis is the elicitation of requirements, possibilities and challenges in order to conceptually design such a system, including data integration, natural language understanding, response generation, and the presentation of analytical results.

  In addition, the paper will identify and critically evaluate the key technical and organizational challenges associated with developing and deploying AI-supported dialogue systems for data analytics. These challenges may include data quality, model reliability, transparency and explainability, user trust, integration into existing organizational workflows, and governance considerations. Possible solution approaches and best practices to address these challenges shall be discussed.

  Literatur

  • Awad, M. N. A., Ivanov, S., Tikhonova, O., & Khodnenko, I. (2025). A Multimodal Conversational Agent for Tabular Data Analysis. arXiv preprint arXiv:2511.18405.
  • Peng, J., Wu, W., Yan, J. N., Qi, D., Rzeszotarski, J. M., & Wang, J. (2022). User Interfaces for Exploratory Data Analysis: A Survey of Open-Source and Commercial Tools. IEEE Data Eng. Bull., 45(3), 116-128.
  • Quamar, A., Efthymiou, V., Lei, C., & Lei, C. (2022). Natural language interfaces to data. Foundations and Trends in Databases, 11(4), 319-414.
  • Stein, C., Teubner, T., & Morana, S. (2024). Designing a conversational agent for supporting data exploration in citizen science. Electronic Markets, 34(1), 23.

As Artificial Intelligence (AI) becomes increasingly embedded in organizational processes, the focus shifts from mere technical feasibility to the question of how AI can be meaningfully integrated into existing work, communication, and decision-making structures (Baird & Maruping, 2021). Today, AI systems support a wide range of organizational tasks, for example by generating analyses, forecasts, recommendations, or communication content. However, their value does not emerge independently of human actors, but rather through the ways in which people interpret, use, evaluate, and communicate AI-supported outputs (Burton et al., 2020; Schilke & Reimann, 2025).

From an IS perspective, it is therefore not only relevant what AI can technically accomplish, but also how individuals respond to AI-supported suggestions and what organizational consequences arise from these responses (Mahmud et al., 2023). In practice, the value of such systems depends to a considerable extent on whether algorithmic recommendations are accepted, adapted, or rejected, and on how transparently the use of AI is communicated in organizational settings (Lim & Schmälzle, 2024; Logg et al., 2019). This topic complex addresses these issues by focusing, first, on the evaluation of algorithmic recommendations in decision-making processes and, second, on the effects of AI disclosure in organizational communication.

Literatur

• Baird, A., & Maruping, L. M. (2021). The next generation of research on IS use: A theoretical framework of delegation to and from agentic IS artifacts. MIS Quarterly, 45(1), 315-341.
• Burton, J. W., Stein, M.-K., & Jensen, T. B. (2020). A systematic review of algorithm aversion in augmented decision making. Journal of Behavioral Decision Making, 33(2), 220-239.
• Lim, S., & Schmälzle, R. (2024). The effect of source disclosure on evaluation of AI-generated messages. Computers in Human Behavior: Artificial Humans, 2, 100058.
• Logg, J. M., Minson, J. A., & Moore, D. A. (2019). Algorithm appreciation: People prefer algorithmic to human judgment. Organizational Behavior and Human Decision Processes, 151, 90-103.
• Mahmud, H., Islam, A. K. M. N., & Mitra, R. K. (2023). What drives managers towards algorithm aversion and how to overcome it? Technological Forecasting and Social Change, 193, 122641.
• Schilke, O., & Reimann, M. (2025). The transparency dilemma: How AI disclosure erodes trust. Organizational Behavior and Human Decision Processes, 188, 104405.

Liste der möglichen konkreten Themen:

• EPA-BA-5-1, Sommersemester 2026, Betreuung: Dr. Christina Strauss

  Between Algorithm Aversion and Appreciation: How People Evaluate AI Recommendations in Organizational Decision-Making – A Systematic Literature Review

  In many organizations, decisions are no longer based solely on human experience and judgment but are increasingly supported by AI systems that autonomously generate forecasts, prioritize information, or provide specific recommendations (Baird & Maruping, 2021). While these systems promise efficiency gains and improved decision quality, research suggests that algorithmic recommendations are not automatically accepted by human decision-makers (Burton et al., 2020). Instead, individuals may react quite differently to AI-generated advice depending on the context and their perceptions of the system.

  Existing literature points to a tension between algorithm aversion and algorithm appreciation. In some situations, people reject algorithmic recommendations even when these perform well, whereas in other contexts they prefer them precisely because they are perceived as objective, consistent, or accurate (Logg et al., 2019; Mahmud et al., 2024). For organizations, this issue is highly relevant because the value of AI-supported decision-making depends on whether employees, managers, or other decision-makers are willing to incorporate such recommendations into their actual decisions. Potential determinants include trust in algorithms, perceived performance, perceived control, task characteristics, familiarity with the technology, and resistance to innovation (Burton et al., 2020; Mahmud et al., 2023).

  The aim of this seminar paper is to conduct a systematic literature review that synthesizes the state of research on algorithm aversion and algorithm appreciation in organizational decision-making contexts. The review should identify the main factors influencing the evaluation of AI-supported recommendations and discuss their implications for organizational decision-making.

  Literatur

  • Burton, J. W., Stein, M.-K., & Jensen, T. B. (2020). A systematic review of algorithm aversion in augmented decision making. Journal of Behavioral Decision Making, 33(2), 220-239.
  • Fisch, C., & Block, J. (2018). Six tips for your (systematic) literature review in business and management research. Management Review Quarterly, 68(2), 103-106.
  • Logg, J. M., Minson, J. A., & Moore, D. A. (2019). Algorithm appreciation: People prefer algorithmic to human judgment. Organizational Behavior and Human Decision Processes, 151, 90-103.
  • Mahmud, H., Islam, A. K. M. N., Luo, X. R., & Mikalef, P. (2024). Decoding algorithm appreciation: Unveiling the impact of familiarity with algorithms, tasks, and algorithm performance. Decision Support Systems, 177, 114168.
  • Mahmud, H., Islam, A. K. M. N., & Mitra, R. K. (2023). What drives managers towards algorithm aversion and how to overcome it? Technological Forecasting and Social Change, 193, 122641.

• EPA-BA-5-2, Sommersemester 2026, Betreuung: Dr. Christina Strauss

  To Disclose or Not to Disclose? Effects of AI Disclosure on the Evaluation of AI-Generated Communication in Organizations - A Systematic Literature Review

  Generative AI is increasingly used in organizations to create or support texts, presentations, reports, marketing content, and other communication formats (Schilke & Reimann, 2025). As a result, a new organizational question arises: Should the use of generative AI in such communication processes be disclosed, and what effects does such disclosure have on how these contents are perceived? 

  Recent research suggests that AI disclosure can function as a signal of transparency but does not necessarily lead to positive evaluations. Depending on the context, disclosing the use of AI may strengthen or weaken trust, credibility, authenticity, and legitimacy (Schilke & Reimann, 2025; Brüns & Meißner, 2024). For organizations, this is particularly relevant because AI-generated communication is used in both internal and external settings and addresses different target groups. Accordingly, disclosure may be evaluated differently in customer communication, marketing, reporting, or other organizational communication processes. At the same time, contextual factors such as the type of content, perceived quality, attitudes toward technology, and expectations of authenticity are likely to shape these effects (Lim & Schmälzle, 2024).

  The aim of this seminar paper is to conduct a systematic literature review that synthesizes research on the effects of AI disclosure on the evaluation of AI-generated communication. The review should synthesize the main factors influencing the evaluation of AI-generated communication and discuss their implications for organizations.

  Literatur

  • Brüns, J. D., & Meißner, M. (2024). Do you create your content yourself? Using generative artificial intelligence for social media content creation diminishes perceived brand authenticity. Journal of Retailing and Consumer Services, 80, 103790.
  • Fisch, C., & Block, J. (2018). Six tips for your (systematic) literature review in business
  • Lim, S., & Schmälzle, R. (2024). The effect of source disclosure on evaluation of AI-generated messages. Computers in Human Behavior: Artificial Humans, 2, 100058.
  • Schilke, O., & Reimann, M. (2025). The transparency dilemma: How AI disclosure erodes trust. Organizational Behavior and Human Decision Processes, 188, 104405.

Generative Artificial Intelligence has rapidly emerged as a transformative technology that can reshape how knowledge is created, processed, and applied across organizations. Systems such as large language models and diffusion models can generate text, images, code, and other content, enabling new forms of automation, creativity, and decision support.

In this topic complex, the focus will not be on the technical development of generative AI models, but rather on their application and implications in research and organizational contexts. Students will explore how generative AI can support research processes and knowledge work, as well as the challenges related to quality, reliability, governance, and ethical use.

Liste der möglichen konkreten Themen:

• SITM-BA-1-1, Sommersemester 2026, Betreuung: Alexandar Schkolski , M.Sc.

  Historical Development of Generative AI and Its Impact on Knowledge Work and Research Processes

  Generative Artificial Intelligence has evolved from early symbolic and rule-based AI systems to advanced models capable of producing text, images, code, and other complex outputs. While early approaches relied on predefined rules and structured logic, later developments in machine learning, neural networks, and transformer architectures enabled data-driven language modeling and large-scale generative capabilities. These advances have significantly expanded the application of generative AI in research and knowledge-intensive environments, supporting content creation, information synthesis, and decision-making processes.

  The student’s task is to examine the historical development and conceptual evolution of generative AI and to analyze its impact on knowledge work and research processes. This includes tracing key milestones from rule-based systems to modern large language models, as well as evaluating how these technological shifts have transformed research workflows, human–AI collaboration, and the organization of intellectual work.

  This paper is designed as a structured literature review. The central research question is: How has generative AI evolved from rule-based systems to modern LLMs? By systematically analyzing academic and relevant conceptual literature, the study should provide a coherent overview of the technological evolution and its implications for research and knowledge-driven activities.

  Literatur

  • Brusco Pletsch, A. L., Tonial, G., Matos, F., & Lauermann Koch, L. (2026). Digital transformation: the role of generative AI in the evolution of knowledge management systems. Journal of Knowledge Management, 1-27.
  • Gopal, R. D., Li, J., Riemer, K., Sarker, S., Singh, P. V., Susarla, A., ... & Thatcher, J. B. (2025). Inventing with machines: Generative ai and the evolving landscape of is research. Information Systems Research, 36(4), 1949-1967.
  • Sowa, K., & Przegalinska, A. (2025). From expert systems to generative artificial experts: a new concept for human-AI collaboration in knowledge work. Journal of Artificial Intelligence Research, 82, 2101-2124.
  • Yoo, Y. (2024). Evolving epistemic infrastructure: The role of scientific journals in the age of generative AI. Journal of the Association for Information Systems, 25(1), 137-144.

• SITM-BA-1-2, Sommersemester 2026, Betreuung: Dr. Erik Karger

  Critical Assessment of Generative AI Integration in Information Systems: Governance, Reliability and Decision Support

  Generative Artificial Intelligence is increasingly embedded in modern information systems, enabling automated content creation, analytical support, code generation, and enhanced human–system interaction. By integrating large language models and related technologies into organizational infrastructures, companies seek to improve efficiency and innovation. However, this integration raises critical concerns regarding reliability, transparency, bias, data protection, and system accountability. In decision-support contexts in particular, questions of trust, governance structures, and validation mechanisms become central, as AI-generated outputs may directly influence managerial and operational decisions.

  The student’s task is to critically examine the integration of generative AI into information systems with a specific focus on governance, reliability, and decision support. This includes analyzing potential benefits and risks, evaluating trust-building mechanisms, and assessing how governance frameworks can ensure responsible and reliable AI deployment. Furthermore, the paper should explore how generative AI affects system design, organizational control structures, and the quality of AI-supported decision-making.

  This paper is designed as a structured literature review. The central research question is: How can generative AI be integrated into information systems in a way that ensures trust, reliability, and effective governance in decision-support contexts? By systematically reviewing academic and relevant practitioner literature, the study should provide a balanced and conceptually grounded assessment of generative AI as a component of modern information systems.

  Literatur

  • Sabherwal, R., & Grover, V. (2024). The societal impacts of generative artificial intelligence: A balanced perspective. Journal of the association for information systems, 25(1), 13-22.
  • Storey, V. C., Yue, W. T., Zhao, J. L., & Lukyanenko, R. (2025). Generative artificial intelligence: Evolving technology, growing societal impact, and opportunities for information systems research. Information Systems Frontiers, 1-22.
  • Van Slyke, C., Johnson, R. D., & Sarabadani, J. (2023). Generative artificial intelligence in information systems education: Challenges, consequences, and responses. Communications of the Association for Information Systems, 53(1), 1-21.
  • Wach, K., Duong, C. D., Ejdys, J., Kazlauskaitė, R., Korzynski, P., Mazurek, G., ... & Ziemba, E. (2023). The dark side of generative artificial intelligence: A critical analysis of controversies and risks of ChatGPT. Entrepreneurial Business and Economics Review, 11(2), 7-30.

• SITM-BA-1-3, Sommersemester 2026, Betreuung: Dr. Erik Karger

  Future Scenarios for Generative AI in Information Systems: Opportunities, Risks, and Strategic Implications

  Generative Artificial Intelligence is expected to significantly influence the future development of information systems. Advances in large language models, multimodal architectures, and increasingly autonomous AI agents suggest that future systems may become more adaptive, context-aware, and capable of supporting complex analytical and decision-making tasks. In organizational settings, generative AI has the potential to transform system architectures, automate knowledge-intensive processes, and redefine human–AI collaboration. At the same time, uncertainties remain regarding scalability, governance, regulatory compliance, risk exposure, and long-term strategic positioning.

  The student’s task is to explore possible future scenarios for the integration of generative AI into information systems. This includes identifying technological trends, analyzing potential application fields, and assessing strategic implications for organizations. The paper should develop structured and plausible future scenarios while critically evaluating associated opportunities, risks, and transformation pathways.

  This paper is designed as a structured literature review with elements of scenario analysis. The central research question is: What plausible future scenarios emerge for the integration of generative AI into information systems, and what strategic implications do they entail for organizations? By systematically reviewing academic research and forward-looking practitioner literature, the study should provide a conceptually grounded assessment of long-term opportunities and risks.

  Literatur

  • Ooi, K. B., Tan, G. W. H., Al-Emran, M., Al-Sharafi, M. A., Capatina, A., Chakraborty, A., ... & Wong, L. W. (2025). The potential of generative artificial intelligence across disciplines: Perspectives and future directions. Journal of Computer Information Systems, 65(1), 76-107.
  • Storey, V. C., Yue, W. T., Zhao, J. L., & Lukyanenko, R. (2025). Generative artificial intelligence: Evolving technology, growing societal impact, and opportunities for information systems research. Information Systems Frontiers, 1-22.
  • Stoykova, S., & Shakev, N. (2023). Artificial intelligence for management information systems: Opportunities, challenges, and future directions. Algorithms, 16(8), 357.
  • Van Slyke, C., Johnson, R. D., & Sarabadani, J. (2023). Generative artificial intelligence in information systems education: Challenges, consequences, and responses. Communications of the Association for Information Systems, 53(1), 1-21.

Large Language Models represent a significant breakthrough in natural language processing, serving as the powerful engines behind modern conversational AI. Built on the transformer architecture and trained on vast datasets, these models have redefined how machines understand, interpret, and generate human-like text across various languages and domains.

In this topic complex, the focus is on exploring the mechanics and optimization of LLMs for specialized research and professional tasks. Students will examine core technical concepts such as tokenization and the mathematical foundations of vector databases, alongside advanced implementation strategies like Retrieval-Augmented Generation (RAG). A central part of the research involves addressing systemic limitations—specifically the phenomenon of hallucinations—and developing mitigation strategies to ensure the accuracy, transparency, and reliable integration of LLMs into data-driven environments.

Liste der möglichen konkreten Themen:

• SITM-BA-2-1, Sommersemester 2026, Betreuung: Alexandar Schkolski , M.Sc.

  Causes of Hallucinations in Large Language Models and Strategies for Mitigation

  Large Language Models, particularly those based on transformer architectures, generate outputs through tokenization and probabilistic next-token prediction. While these models demonstrate strong linguistic fluency and contextual understanding, they are prone to so-called “hallucinations,” producing statements that appear coherent but are factually incorrect, inconsistent, or fabricated. Such behavior is rooted in the statistical nature of language modeling, limitations in training data, model architecture constraints, and the absence of true semantic understanding. In research and decision-support contexts, these inaccuracies pose risks to reliability, trustworthiness, and informed decision-making.

  The student’s task is to examine the technical and conceptual causes of hallucinations in LLMs and to analyze existing mitigation strategies. This includes investigating tokenization mechanisms, transformer-based prediction processes, and data-related influences, as well as evaluating approaches such as prompt engineering, retrieval-augmented generation, fine-tuning, and external validation mechanisms. The paper should assess both underlying causes and practical countermeasures in a structured and analytical manner.

  This paper is designed as a structured literature review. The central research question is: What are the primary causes of hallucinations in large language models, and which mitigation strategies are most effective in addressing them? By systematically reviewing academic and practitioner-oriented literature, the study should provide a conceptually grounded overview of technical challenges and solution approaches.

  Literatur

  • Abdelghafour, M. A. M., Mabrouk, M., & Taha, Z. (2024). Hallucination mitigation techniques in large language models. International Journal of Intelligent Computing and Information Sciences, 24(4), 73-81.
  • Ahmadi, A. (2024). Unravelling the mysteries of hallucination in large language models: strategies for precision in artificial intelligence language generation. Asian Journal of Computer Science and Technology, 13(1), 1-10.
  • Ciubotaru, B. I. (2025). The hallucination problem in generative artificial intelligence: Accuracy and trust in digital learning. In International Conference on Virtual Learning (Vol. 20, pp. 35-45).
  • Das, S., Chatterji, S., & Mukherjee, I. (2023, December). Combating Hallucination and Misinformation: Factual Information Generation with Tokenized Generative Transformer. In Proceedings of the Joint 3rd International Conference on Natural Language Processing for Digital Humanities and 8th International Workshop on Computational Linguistics for Uralic Languages (pp. 143-152).

• SITM-BA-2-2, Sommersemester 2026, Betreuung: Alexandar Schkolski , M.Sc.

  Vector Databases and Retrieval-Augmented Generation: Architecture, Applications, and Case Studies

  Vector databases have become a key infrastructure component in modern generative AI systems, particularly in applications requiring semantic search and context-aware information retrieval. Instead of storing data in relational tables, vector databases manage high-dimensional embeddings that represent textual, visual, or multimodal data within a mathematical vector space. These embeddings, generated by neural networks, enable similarity-based retrieval using distance metrics such as cosine similarity or Euclidean distance. In Retrieval-Augmented Generation (RAG) architectures, vector databases provide external knowledge to large language models, thereby improving factual grounding, contextual relevance, and response reliability.

  The student’s task is to examine the architectural and mathematical foundations of vector databases and their role within Retrieval-Augmented Generation systems. This includes explaining embeddings, vector spaces, similarity measures, indexing mechanisms, and the integration of retrieval pipelines into generative workflows. Furthermore, the paper should analyze practical applications and case studies in which RAG architectures enhance performance, transparency, and trustworthiness of AI-generated outputs.

  This paper is designed as a structured literature review with illustrative case analysis. The central research question is: How do vector databases and Retrieval-Augmented Generation architectures enhance the reliability and performance of generative AI systems? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded understanding of vector databases as an enabling technology for knowledge-augmented AI applications.

  Literatur

  • Akik, E., Vještica, M., Tomić, M., Slivka, J., Čeliković, M., & Kordić, S. (2025, March). Plagiarism Detection of Student Assignments: The Application of Retrieval-Augmented Generation and Vector Database. In Conference on Information Technology and its Applications (pp. 303-319). Cham: Springer Nature Switzerland.
  • Joshi, S. (2025). Introduction to vector databases for generative AI: Applications, performance, future projections, and cost considerations. International Advanced Research Journal in Science, Engineering and Technology ISSN (O), 2393-8021.
  • KASA HALILI, M., HALILI, F., KADRIU, S., & MAZLAMI, V. (2025). THE ROLE OF VECTOR DATABASES IN THE ERA OF GENERATIVE AI. Journal of Applied Sciences-SUT, 11(21-22), 174-183.
  • Rusum, G. P., & Anasuri, S. (2024). Vector Databases in Modern Applications: Real-Time Search, Recommendations, and Retrieval-Augmented Generation (RAG). International Journal of AI, BigData, Computational and Management Studies, 5(4), 124-136.

• SITM-BA-2-3, Sommersemester 2026, Betreuung: Deniz Baris Gölgelioglu , M. Sc.

  Architecture and Training of Large Language Models: Processes, Data, and System Design

  Large Language Models (LLMs) are based on transformer architectures and require complex, multi-stage training processes to achieve their generative capabilities. Key stages include large-scale unsupervised pretraining on diverse datasets, fine-tuning for specific tasks, and alignment procedures to ensure outputs meet desired quality, safety, and ethical standards. Scaling principles, including model size, number of parameters, and computational resources, critically influence performance, generalization, and reasoning ability. The interaction between architectural design, training data, and system infrastructure forms a sophisticated ecosystem that determines model effectiveness, reliability, and scalability.

  The student’s task is to examine the architecture and training processes of LLMs with particular emphasis on pretraining strategies, alignment methods, and scaling considerations. This includes analyzing dataset curation, preprocessing approaches, model parameter scaling, and fine-tuning techniques, as well as their impact on performance, bias, and domain adaptation. The paper should provide a structured overview of how architectural design, data selection, and training procedures interact, and critically reflect on challenges such as resource consumption, ethical implications, and alignment with intended applications.

  This paper is designed as a structured literature review. The central research question is: How do pretraining, alignment, and scaling strategies influence the architecture, performance, and reliability of large language models? By systematically reviewing academic and practitioner literature, the study should provide a comprehensive understanding of LLM training processes and design considerations.

  Literatur

  • Chang, Y., Wang, X., Wang, J., Wu, Y., Yang, L., Zhu, K., ... & Xie, X. (2024). A survey on evaluation of large language models. ACM transactions on intelligent systems and technology, 15(3), 1-45.
  • Jiang, Z., Lin, H., Zhong, Y., Huang, Q., Chen, Y., Zhang, Z., ... & Liu, X. (2024). {MegaScale}: Scaling large language model training to more than 10,000 {GPUs}. In 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24) (pp. 745-760).
  • Naveed, H., Khan, A. U., Qiu, S., Saqib, M., Anwar, S., Usman, M., ... & Mian, A. (2025). A comprehensive overview of large language models. ACM Transactions on Intelligent Systems and Technology, 16(5), 1-72.
  • Raiaan, M. A. K., Mukta, M. S. H., Fatema, K., Fahad, N. M., Sakib, S., Mim, M. M. J., ... & Azam, S. (2024). A review on large language models: Architectures, applications, taxonomies, open issues and challenges. IEEE access, 12, 26839-26874.

• SITM-BA-2-4, Sommersemester 2026, Betreuung: Deniz Baris Gölgelioglu , M. Sc.

  Deployment and Integration of Large Language Models in Information Systems

  Large Language Models (LLMs) are increasingly deployed via application interfaces (APIs) and integrated into diverse information systems to provide generative functionalities such as automated text generation, summarization, translation, and conversational support. Deployment architectures typically include API endpoints, cloud-based infrastructure, authentication layers, and orchestration of input-output workflows. Effective integration requires careful consideration of latency, scalability, data flow, security, and interoperability with existing software components and organizational processes. Proper deployment ensures that LLMs can reliably support decision-making, knowledge work, and operational efficiency.

  The student’s task is to analyze the deployment and integration of LLMs within information systems, focusing on system architecture, APIs, and practical implementation considerations. This includes examining infrastructure requirements, interaction patterns, performance optimization, security measures, and strategies for maintaining reliability and scalability. The paper should provide a structured overview of how LLMs are operationalized in real-world research, business, and knowledge-intensive applications.

  This paper is designed as a structured literature review. The central research question is: How can large language models be effectively deployed and integrated into information systems to ensure reliable, secure, and scalable operation? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded understanding of deployment architectures, integration strategies, and operational considerations.

  Literatur

  • Annepaka, Y., & Pakray, P. (2025). Large language models: a survey of their development, capabilities, and applications. Knowledge and Information Systems, 67(3), 2967-3022.
  • Han, S., Wang, M., Zhang, J., Li, D., & Duan, J. (2024). A review of large language models: Fundamental architectures, key technological evolutions, interdisciplinary technologies integration, optimization and compression techniques, applications, and challenges. Electronics, 13(24), 5040.
  • Pesl, R. D. (2025, June). Adopting large language models to automated system integration. In International Conference on Advanced Information Systems Engineering (pp. 313-320). Cham: Springer Nature Switzerland.
  • Raiaan, M. A. K., Mukta, M. S. H., Fatema, K., Fahad, N. M., Sakib, S., Mim, M. M. J., ... & Azam, S. (2024). A review on large language models: Architectures, applications, taxonomies, open issues and challenges. IEEE access, 12, 26839-26874.

• SITM-BA-2-5, Sommersemester 2026, Betreuung: Deniz Baris Gölgelioglu , M. Sc.

  Systems for Using Large Language Models: Evaluation and Comparison of Tools like LM Studio

  A growing number of platforms and systems facilitate the use of Large Language Models (LLMs), such as LM Studio and similar tools, enabling users to deploy, experiment with, and customize generative AI without requiring deep technical expertise. These platforms vary in capabilities, user interfaces, supported models, integration options, scalability, and performance. Effective evaluation requires examining usability, computational efficiency, model support, security, and compliance. Additionally, comparing local versus cloud-based solutions is critical, particularly with respect to data privacy, latency, and control over sensitive information. Understanding these factors is essential for selecting the most suitable tool for research, organizational workflows, or development tasks.

  The student’s task is to analyze and compare systems for using LLMs, including LM Studio and alternative platforms. This includes developing evaluation criteria, assessing features and limitations, and performing structured comparisons that account for deployment modes (local vs. cloud), privacy considerations, and practical use cases. The paper should provide a framework for assessing generative AI platforms, highlighting their strengths, weaknesses, and suitability for different organizational or research contexts.

  This paper is designed as a structured literature review and comparative analysis. The central research question is: How do different LLM platforms, including local and cloud-based solutions, compare in terms of functionality, usability, performance, and data privacy? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded framework for evaluating and selecting LLM platforms.

  Literatur

  • Chang, Y., Wang, X., Wang, J., Wu, Y., Yang, L., Zhu, K., ... & Xie, X. (2024). A survey on evaluation of large language models. ACM transactions on intelligent systems and technology, 15(3), 1-45.
  • Hadi, M. U., Qureshi, R., Shah, A., Irfan, M., Zafar, A., Shaikh, M. B., ... & Mirjalili, S. (2023). Large language models: a comprehensive survey of its applications, challenges, limitations, and future prospects. Authorea preprints, 1(3), 1-26.
  • Hou, X., Zhao, Y., Liu, Y., Yang, Z., Wang, K., Li, L., ... & Wang, H. (2024). Large language models for software engineering: A systematic literature review. ACM Transactions on Software Engineering and Methodology, 33(8), 1-79.
  • Xu, F. F., Alon, U., Neubig, G., & Hellendoorn, V. J. (2022, June). A systematic evaluation of large language models of code. In Proceedings of the 6th ACM SIGPLAN international symposium on machine programming (pp. 1-10).

• SITM-BA-2-6, Sommersemester 2026, Betreuung: Alexandar Schkolski , M.Sc.

  Different Architectures for Personal Information Management: Cloud, Local, and Hybrid Approaches

  Personal Information Management (PIM) systems assist users in organizing, retrieving, and interacting with personal data, including notes, documents, schedules, and communications. Modern PIM architectures differ widely in design and deployment. Cloud-based solutions synchronize data across devices and support collaborative workflows, local solutions store data on-device for enhanced privacy and offline access, and hybrid architectures combine cloud and local storage to balance accessibility, performance, and security. Each architecture has distinct implications for usability, data protection, accessibility, and system performance, shaping how individuals manage and leverage their personal information.

  The student’s task is to analyze and compare different architectures for personal information management, focusing on cloud, local, and hybrid approaches. This includes evaluating usability, data management strategies, privacy and security considerations, and integration capabilities. The paper should provide a structured assessment of how these architectures support personal information workflows while highlighting the strengths, limitations, and trade-offs of each approach.

  This paper is designed as a structured literature review and comparative analysis. The central research question is: How do cloud, local, and hybrid PIM architectures differ in supporting usability, security, and data management, and what are their respective advantages and limitations? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded framework for evaluating PIM architectures.

  Literatur

  • Adavi, K. A. K., & Acker, A. (2023). What is a file on a phone? Personal information management practices amongst WhatsApp users. Proceedings of the ACM on Human-Computer Interaction, 7(CSCW2), 1-28.
  • Al Nasar, M. R., Mohd, M., & Ali, N. M. (2011, June). Personal information management systems and interfaces: An overview. In 2011 International Conference on Semantic Technology and Information Retrieval (pp. 197-202). IEEE.
  • Jones, W. (2022). Transforming technologies to manage our information: The future of personal information management, Part 2. Springer Nature.
  • Wang, S. W., & Chang, S. E. (2014, April). Personal data management: an architectural framework for personal cloud mobile application. In 2014 International Conference on Information Science, Electronics and Electrical Engineering (Vol. 2, pp. 781-785). IEEE.

• SITM-BA-2-7, Sommersemester 2026, Betreuung: Dr. Erik Karger

  Overview of Large Language Models and Their Support for the Research Process

  A wide range of Large Language Models (LLMs) is available today, including open-source solutions, commercial offerings, and specialized domain-specific models. These models vary in size, architecture, training data, capabilities, and access methods, which influence their suitability for different research tasks. LLMs can support the research process by assisting with literature review, data analysis, summarization, hypothesis generation, and drafting of academic content. They also enable novel forms of knowledge exploration, collaboration, and workflow automation, while raising important considerations regarding reliability, bias, and ethical use.

  The student’s task is to provide an overview of existing LLMs and analyze how they can support various stages of the research process. This includes comparing models in terms of capabilities, accessibility, and performance, and evaluating their practical applications in research workflows. The paper should offer a structured assessment of the strengths, limitations, and potential of LLMs as tools for enhancing research efficiency, quality, and creativity.

  This paper is designed as a structured literature review. The central research question is: How do different Large Language Models support various stages of the research process, and what are their respective strengths, limitations, and practical applications? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded understanding of LLMs in research contexts.

  Literatur

  • Baek, J., Jauhar, S. K., Cucerzan, S., & Hwang, S. J. (2025). Researchagent: Iterative research idea generation over scientific literature with large language models. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) (pp. 6709-6738).
  • Jansen, B. J., Jung, S. G., & Salminen, J. (2023). Employing large language models in survey research. Natural Language Processing Journal, 4, 100020.
  • Liu, Y., Han, T., Ma, S., Zhang, J., Yang, Y., Tian, J., ... & Ge, B. (2023). Summary of chatgpt-related research and perspective towards the future of large language models. Meta-radiology, 1(2), 100017.
  • Naveed, H., Khan, A. U., Qiu, S., Saqib, M., Anwar, S., Usman, M., ... & Mian, A. (2025). A comprehensive overview of large language models. ACM Transactions on Intelligent Systems and Technology, 16(5), 1-72.

• SITM-BA-2-8, Sommersemester 2026, Betreuung: Alexandar Schkolski , M.Sc.

  Large Language Models for Software Development: Approaches, Advantages, and Limitations

  Large Language Models (LLMs) are increasingly applied in software development to support code generation, debugging, documentation, and problem-solving. Various approaches exist, including general-purpose code models, models fine-tuned for specific programming languages, and integration via IDE plugins or APIs. These tools can accelerate development, improve code quality, and support learning, while also introducing challenges such as generating incorrect or insecure code, limited awareness of project context, and potential over-reliance on automated suggestions.

  The student’s task is to analyze the application of LLMs in software development, examining different approaches, their advantages, and their limitations. This includes evaluating model effectiveness, integration strategies, and practical and ethical risks. The paper should provide a structured assessment of how LLMs can enhance development workflows while critically reflecting on potential pitfalls, safety considerations, and responsible use.

  This paper is designed as a structured literature review. The central research question is: How can Large Language Models be effectively applied in software development, and what are the advantages, limitations, and risks of different approaches? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded evaluation of LLMs as tools for software engineering.

  Literatur

  • Fan, A., Gokkaya, B., Harman, M., Lyubarskiy, M., Sengupta, S., Yoo, S., & Zhang, J. M. (2023, May). Large language models for software engineering: Survey and open problems. In 2023 IEEE/ACM International Conference on Software Engineering: Future of Software Engineering (ICSE-FoSE) (pp. 31-53). IEEE.
  • Hou, X., Zhao, Y., Liu, Y., Yang, Z., Wang, K., Li, L., ... & Wang, H. (2024). Large language models for software engineering: A systematic literature review. ACM Transactions on Software Engineering and Methodology, 33(8), 1-79.
  • Ozkaya, I. (2023). Application of large language models to software engineering tasks: Opportunities, risks, and implications. IEEE software, 40(3), 4-8.
  • Zheng, Z., Ning, K., Zhong, Q., Chen, J., Chen, W., Guo, L., ... & Wang, Y. (2025). Towards an understanding of large language models in software engineering tasks. Empirical Software Engineering, 30(2), 50.

• SITM-BA-2-9, Sommersemester 2026, Betreuung: Deniz Baris Gölgelioglu , M. Sc.

  Large Language Models for Business Process and Enterprise Modeling: Preconditions, Approaches, and Comparative Analysis

  Large Language Models (LLMs) are increasingly applied in business process management (BPM) and enterprise modeling to simulate, analyze, and optimize organizational structures, workflows, and decision processes. Effective application of LLMs requires specific preconditions, including access to structured organizational data, clearly defined modeling objectives, and integration with analytical and BPM frameworks. Different approaches exist, ranging from using LLMs for scenario generation, process simulation, and workflow optimization to embedding them within decision-support systems or knowledge management platforms. Each approach differs in scalability, interpretability, accuracy, and alignment with enterprise goals and operational realities.

  The student’s task is to examine the use of LLMs in business process and enterprise modeling, analyzing preconditions for effective implementation, comparing different methodological approaches, and evaluating their respective strengths and limitations. The paper should provide a structured assessment of how LLMs can support BPM, enterprise structure analysis, and informed decision-making, while highlighting practical, technical, and conceptual considerations.

  This paper is designed as a structured literature review and comparative analysis. The central research question is: How can Large Language Models be effectively applied in business process and enterprise modeling, and what are the preconditions, approaches, and limitations of different methods? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded understanding of LLMs as tools for enterprise and process modeling.

  Literatur

  • Berezovsky, V., Sluzova, N., Vahrushev, A., Gubkin, D., & Potyomin, V. (2025, July). Intelligent Support and Modeling of Organizational Structure Redesign for Micro and Small Businesses based on Large Language Models and Fuzzy Rules. In International Conference on Intelligent and Fuzzy Systems (pp. 535-543). Cham: Springer Nature Switzerland.
  • Görgen, L., Müller, E., Triller, M., Nast, B., & Sandkuhl, K. (2024). Large Language Models in Enterprise Modeling: Case Study and Experiences. In MODELSWARD (pp. 74-85).
  • Kourani, H., Berti, A., Schuster, D., & van der Aalst, W. M. (2024, May). Process modeling with large language models. In International Conference on Business Process Modeling, Development and Support (pp. 229-244). Cham: Springer Nature Switzerland.
  • Ziche, C., & Apruzzese, G. (2024, September). LLM4PM: A case study on using large language models for process modeling in enterprise organizations. In International conference on business process management (pp. 472-483). Cham: Springer Nature Switzerland.

• SITM-BA-2-10, Sommersemester 2026, Betreuung: Falco Korn , M.Sc.

  Large Language Models for User Interface Generation: Languages, Current Status, and Use Cases

  Large Language Models (LLMs) are increasingly used to generate user interfaces (UIs), enabling automated creation of layouts, code snippets, and interaction flows from natural language descriptions. These models can generate code for multiple programming and markup languages, including HTML, CSS, JavaScript, and modern frameworks such as React, Angular, Flutter, and Vue.js. While still an emerging technology, LLM-driven UI generation offers opportunities to accelerate prototyping, improve accessibility, and assist developers in creating responsive, adaptive, and context-aware interfaces. Current use cases include rapid UI prototyping, generating dynamic dashboards, and adapting interface components to user preferences or workflow requirements.

  The student’s task is to examine the application of LLMs in user interface generation, analyzing the supported programming languages and frameworks, evaluating the current technological state, and exploring practical use cases. The paper should assess strengths, limitations, and potential applications, providing a structured overview of how LLMs can enhance UI design, development efficiency, and accessibility in real-world projects.

  This paper is designed as a structured literature review with applied examples. The central research question is: How can Large Language Models be used to generate user interfaces across different languages and frameworks, and what are the opportunities and limitations of current approaches? By systematically reviewing academic literature and industry examples, the study should provide a conceptually grounded understanding of LLM-enabled UI generation.

  Literatur

  • Brade, S., Wang, B., Sousa, M., Oore, S., & Grossman, T. (2023, October). Promptify: Text-to-image generation through interactive prompt exploration with large language models. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology (pp. 1-14).
  • Duan, P., Warner, J., Li, Y., & Hartmann, B. (2024, May). Generating automatic feedback on ui mockups with large language models. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (pp. 1-20).
  • Vaithilingam, P., Zhang, T., & Glassman, E. L. (2022, April). Expectation vs. experience: Evaluating the usability of code generation tools powered by large language models. In Chi conference on human factors in computing systems extended abstracts (pp. 1-7).
  • Wu, J., Schoop, E., Leung, A., Barik, T., Bigham, J. P., & Nichols, J. (2024, June). Uicoder: Finetuning large language models to generate user interface code through automated feedback. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) (pp. 7511-7525). 

• SITM-BA-2-11, Sommersemester 2026, Betreuung: Falco Korn , M.Sc.

  Conversational Interfaces vs Graphical Interfaces in the Age of LLMs

  The emergence of Large Language Models (LLMs) is poised to significantly transform how users interact with applications, potentially shifting interaction paradigms from traditional graphical user interfaces (GUIs) toward conversational, multimodal, or AI-driven interfaces. LLMs can enable adaptive interactions, natural language navigation, and highly personalized user experiences. This shift raises questions about the persistence, evolution, or possible obsolescence of conventional GUIs. Early research and “gray literature,” including industry white papers, design prototypes, and experimental studies, highlight diverse possibilities while also revealing challenges related to usability, accessibility, trust, and user acceptance.

  The student’s task is to explore the evolving landscape of application interfaces in the context of LLM integration. This includes analyzing potential transformations, comparing conversational and graphical interface paradigms, and reviewing gray literature to capture experimental insights and industry trends. The paper should provide a structured assessment of how generative AI may reshape user interaction, interface design, and overall user experience.

  This paper is designed as a structured literature review with comparative analysis. The central research question is: How will the rise of Large Language Models influence the balance between conversational and graphical interfaces, and what are the implications for usability, accessibility, and user experience? By systematically reviewing academic and practitioner sources, the study should provide a conceptually grounded understanding of future interface paradigms.

  Literatur

  • Chen, J., Liu, Z., Huang, X., Wu, C., Liu, Q., Jiang, G., ... & Chen, E. (2024). When large language models meet personalization: Perspectives of challenges and opportunities. World wide web, 27(4), 42.
  • Hadi, M. U., Qureshi, R., Shah, A., Irfan, M., Zafar, A., Shaikh, M. B., ... & Mirjalili, S. (2023). Large language models: a comprehensive survey of its applications, challenges, limitations, and future prospects. Authorea preprints, 1(3), 1-26.
  • Jablonka, K. M., Ai, Q., Al-Feghali, A., Badhwar, S., Bocarsly, J. D., Bran, A. M., ... & Blaiszik, B. (2023). 14 examples of how LLMs can transform materials science and chemistry: a reflection on a large language model hackathon. Digital discovery, 2(5), 1233-1250.
  • Monteiro, M., Branco, B. C., Silvestre, S., Avelino, G., & Valente, M. T. (2025). NoCodeGPT: A No‐Code Interface for Building Web Apps With Language Models. Software: Practice and Experience, 55(8), 1408-1424.

• SITM-BA-2-12, Sommersemester 2026, Betreuung: Falco Korn , M.Sc.

  Large Language Models of Varying Complexity: Capabilities, Reasoning, and Resource Requirements

  Large Language Models (LLMs) vary significantly in size, architecture, and computational demands, which directly affects their capabilities. Smaller models can handle basic text generation and simple reasoning tasks but often struggle with complex problem-solving, long-context understanding, or nuanced outputs. Larger models provide more advanced reasoning, broader knowledge coverage, and improved contextual understanding, but they require substantial computational resources, including high CPU/GPU usage, memory, and storage. Benchmarking—through task-specific evaluations, standardized performance tests, and comparison metrics—helps quantify these differences and informs model selection for research and application contexts.

  The student’s task is to analyze LLMs of differing complexity, comparing their capabilities, reasoning potential, and computational requirements. This includes examining CPU/GPU usage, memory and storage needs, and performance benchmarks, alongside practical examples illustrating small versus large model behavior. The paper should provide a structured assessment of the trade-offs between model size, resource demands, and functional performance, highlighting how benchmarking can guide informed decisions about LLM selection and deployment.

  This paper is designed as a structured literature review with comparative analysis. The central research question is: How do LLMs of varying complexity differ in capabilities, reasoning, and computational requirements, and how can benchmarking guide their effective selection and application? By systematically reviewing academic and practitioner literature, the study should provide a conceptually grounded understanding of performance trade-offs across LLM sizes.

  Literatur

  • Diao, S., Wang, P., Lin, Y., Pan, R., Liu, X., & Zhang, T. (2024, August). Active prompting with chain-of-thought for large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (pp. 1330-1350).
  • Fan, L., Hua, W., Li, L., Ling, H., & Zhang, Y. (2024, August). Nphardeval: Dynamic benchmark on reasoning ability of large language models via complexity classes. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (pp. 4092-4114).
  • He, Q., Zeng, J., Huang, W., Chen, L., Xiao, J., He, Q., ... & Xiao, Y. (2024, March). Can large language models understand real-world complex instructions?. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 38, No. 16, pp. 18188-18196).
  • Jin, M., Yu, Q., Huang, J., Zeng, Q., Wang, Z., Hua, W., ... & Zhang, Y. (2025, January). Exploring concept depth: How large language models acquire knowledge and concept at different layers?. In Proceedings of the 31st international conference on computational linguistics (pp. 558-573).

Die Pflege steht in vielen Ländern vor großen Herausforderungen. Der demografische Wandel führt zu einer steigenden Zahl pflegebedürftiger Menschen, während gleichzeitig ein zunehmender Fachkräftemangel im Pflegebereich zu beobachten ist (Destatis 22026). Zu betonen ist jedoch, dass nicht nur alte Menschen Pflegebedarf haben, sondern durch Krankheiten oder Unfälle es Menschen in allen Altersklassen treffen kann. Aufgrund des hohen Bedarfs stehen Pflegekräfte unter hohem Zeitdruck und müssen eine Vielzahl komplexer Aufgaben bewältigen, von medizinischer Versorgung über Dokumentation bis hin zur emotionalen Betreuung von Patientinnen und Patienten.

Technologische Innovationen, insbesondere im Bereich der künstlichen Intelligenz (KI), generativen KI-Systemen (GenKI) und robotischen Assistenzsystemen, werden zunehmend als mögliche Unterstützung für Pflegekräfte diskutiert. Beispiele reichen von intelligenten Dokumentationssystemen über Assistenzroboter bis hin zu KI-basierten Entscheidungsunterstützungssystemen oder automatisierten Monitoringlösungen. Dennoch ist bislang nur begrenzt verstanden, in welchen konkreten Situationen solche Technologien tatsächlich einen Mehrwert bieten und welche Anforderungen Pflegebedürftige und Pflegekräfte an zukünftige Technologien haben.

Im Rahmen dieses Seminarprojekts untersuchen Studierende unterschiedliche Bereiche der Pflegepraxis. Ziel ist es, durch qualitative Experteninterviews mit Pflegekräften ein besseres Verständnis für typische Arbeitsabläufe, zentrale Herausforderungen sowie den aktuellen Einsatz von Technologien in der Pflege zu gewinnen. Darüber hinaus sollen Erwartungen und Anforderungen an zukünftige KI- und Robotiklösungen identifiziert werden sowie Grenzen technischer Unterstützung diskutiert werden, insbesondere in Bereichen, in denen menschliche Interaktion unverzichtbar bleibt.

Die Ergebnisse der Seminararbeiten sollen dazu beitragen, praxisnahe Einblicke in die Arbeitsrealität der Pflege zu gewinnen und Potenziale für den sinnvollen Einsatz zukünftiger KI- und Robotiklösungen im Pflegekontext zu identifizieren.

Literatur

• Statistisches Bundesamt. (2026). Pflege. https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Gesundheit/Pflege/_inhalt.html
• Gerling, K., Hebesberger, D., Dondrup, C., Körtner, T., & Hanheide, M. (2016). Robot deployment in long-term care. Zeitschrift Fur Gerontologie Und Geriatrie, 49, 288–297. https://doi.org/10.1007/s00391-016-1065-6
• Bratan, T., Schneider, D., Funer, F., Heyen, N. B., Klausen, A., Liedtke, W., Lipprandt, M., Salloch, S., & Langanke, M. (2024). Unterstützung ärztlicher und pflegerischer Tätigkeit durch KI: Handlungsempfehlungen für eine verantwortbare Gestaltung und Nutzung. Bundesgesundheitsblatt, Gesundheitsforschung, Gesundheitsschutz, 67(9), 1039–1046. https://doi.org/10.1007/s00103-024-03918-1
• Zöllick, J. C., Rössle, S., Kluy, L., Kuhlmey, A., & Blüher, S. (2022). Potenziale und Herausforderungen von sozialen Robotern für Beziehungen älterer Menschen: Eine Bestandsaufnahme mittels „rapid review“. Zeitschrift Fur Gerontologie Und Geriatrie, 55(4), 298–304. https://doi.org/10.1007/s00391-021-01932-5
• Laufer, J., Banh, L., & Strobel, G. (2025). Bridging Mind and Matter: A Taxonomy of Embodied Generative AI. In AIS (Ed.), Wirtschaftsinformatik 2025 Proceedings.
• Strobel, G., Banh, L., Möller, F., & Schoormann, T. (2024). Exploring Generative Artificial Intelligence: A Taxonomy and Types. Hawaii International Conference on System Sciences. https://doi.org/10.24251/HICSS.2023.546
• Banh, L., & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets, 33(1), 63. https://doi.org/10.1007/s12525-023-00680-1
• Tuunanen, T., Winter, R., & Brocke, J. V. (2024). Dealing with Complexity in Design Science Research: A Methodology Using Design Echelons. MIS Quarterly, 48(2), 427–458. https://doi.org/10.25300/MISQ/2023/16700
• Myers, M. D., & Newman, M. (2007). The qualitative interview in IS research: Examining the craft. Information and Organization, 17(1), 2–26. https://doi.org/10.1016/j.infoandorg.2006.11.001
• Birks, D. F., Fernandez, W., Levina, N., & Nasirin, S. (2013). Grounded theory method in information systems research: Its nature, diversity and opportunities. European Journal of Information Systems, 22(1), 1–8. https://doi.org/10.1057/ejis.2012.48

Liste der möglichen konkreten Themen:

• SOFTEC-BA-1-1, Sommersemester 2026, Betreuung: Jan Laufer , M. Sc.

  House of Care – KI und Robotik in der stationären Pflege

  Die stationäre Pflege und insbesondere die Altenpflege stellt einen zentralen Bestandteil der Pflegeversorgung dar und gewinnt im Zuge des demografischen Wandels zunehmend an Bedeutung. Pflegekräfte in stationären Pflegeeinrichtungen übernehmen eine Vielzahl an Aufgaben, darunter körperliche Pflege, medizinische Unterstützung, organisatorische Tätigkeiten sowie soziale Betreuung der Bewohnerinnen und Bewohner. Gleichzeitig stehen Pflegeeinrichtungen häufig unter erheblichem personellen und organisatorischen Druck.

  Vor diesem Hintergrund wird zunehmend diskutiert, inwiefern digitale Technologien, künstliche Intelligenz und robotische Assistenzsysteme Pflegekräfte im Alltag unterstützen können. Beispiele reichen von digitalen Dokumentationssystemen über Monitoringlösungen bis hin zu Assistenzrobotern. Gleichzeitig stellt sich die Frage, welche Aufgaben sinnvoll technologisch unterstützt werden können und wo menschliche Betreuung und Interaktion unverzichtbar bleiben. Diese Seminararbeit untersucht die stationäre Pflege aus der Perspektive der Praxis.

  Forschungsfragen: 

  • Welche typischen Arbeitsabläufe und Aufgaben prägen den Alltag von Pflegekräften in der Altenpflege?
  • Welche zentralen Herausforderungen und Belastungen erleben Pflegekräfte im Arbeitsalltag?
  • Welche Technologien werden aktuell eingesetzt und wie werden diese bewertet?
  • Welche Anforderungen und Wünsche haben Pflegekräfte an zukünftige Technologien, insbesondere im Bereich KI und Robotik?
  • Wo sehen Pflegekräfte klare Grenzen für den Einsatz von Technologie in der Pflege?

  Methodik:

  Die Arbeit folgt einem qualitativen Forschungsansatz und orientiert sich konzeptionell am echeloned Design Science Research (eDSR) Ansatz nach Tuunanen et al. (2024). Der eDSR-Ansatz strukturiert Design-Science-Forschung entlang mehrerer sogenannter Design Echelons, die unterschiedliche Ebenen der Problem- und Lösungsentwicklung adressieren. Während höhere Echelons typischerweise die Entwicklung, Implementierung und Evaluation konkreter Artefakte umfassen, konzentriert sich diese Seminararbeit auf die ersten beiden Echelons des Ansatzes. Die weiteren Echelons des eDSR-Ansatzes, die sich beispielsweise mit der konkreten Entwicklung, Implementierung oder Evaluation von Artefakten befassen, sind nicht Gegenstand dieser Seminararbeit und daher bewusst out of scope.

  Echelon 1 – Problem Exploration: In dieser Phase wird das Problemfeld systematisch untersucht, um ein fundiertes Verständnis der realen Praxisprobleme zu entwickeln. Ziel ist es, zentrale Herausforderungen, Arbeitsabläufe sowie bestehende Praktiken im Kontext der stationären Pflege zu identifizieren.

  Echelon 2 – Derivation of Design Knowledge: Auf Grundlage der identifizierten Problemfelder werden erste Designziele und Designanforderungen für potenzielle technologische Unterstützungssysteme abgeleitet.

  Zur Datensammlung werden qualitative Experteninterviews mit Pflegekräften aus dem jeweiligen Pflegekontext durchgeführt. Ziel der Interviews ist es, ein vertieftes Verständnis typischer Arbeitsabläufe, zentraler Herausforderungen sowie des aktuellen Einsatzes von Technologien im Pflegealltag zu gewinnen. Die Gestaltung und Durchführung der Interviews orientiert sich an etablierten Empfehlungen für qualitative Interviewforschung in der Informationssystemforschung (Myers & Newman, 2007).

  Zur Datenauswertung werden die Interviews transkribiert und anschließend mittels induktivem Open Coding analysiert. Ziel dieser Auswertung ist die systematische Identifikation zentraler Problemfelder sowie relevanter Anforderungen aus der Praxis. Die Analyse orientiert sich methodisch an der Grounded Theory Methodology, insbesondere am offenen, induktiven Kodieren qualitativer Daten (Birks et al., 2013).

  Erwarteter Beitrag:

  Die Arbeit liefert praxisnahe Einblicke in die Arbeitsrealität der stationären Pflege und identifiziert Potenziale sowie Grenzen für den Einsatz zukünftiger Technologien in diesem Bereich. Durch die Experteninterviews entsteht zunächst eine strukturierte und möglichst umfassende Übersicht über zentrale Probleme, Herausforderungen und Belastungen im Arbeitsalltag von Pflegekräften in stationären Pflegeeinrichtungen.

  Auf Basis dieser Problemübersicht sollen anschließend konkrete Gestaltungsansätze für zukünftige technologische Unterstützungssysteme abgeleitet werden. Dazu gehören insbesondere die Formulierung von Designzielen sowie Designanforderungen für mögliche Technologien, beispielsweise im Bereich digitaler Dokumentationssysteme, KI-gestützter Assistenzsysteme oder robotischer Unterstützung in Pflegeeinrichtungen.

  Dabei wird ausdrücklich berücksichtigt, dass nicht alle identifizierten Probleme durch Technologie adressiert werden können. Ein wichtiger Teil der Analyse besteht daher darin zu unterscheiden, welche Herausforderungen prinzipiell technologisch unterstützbar sind und in welchen Bereichen menschliche Interaktion, Empathie und persönliche Betreuung unverzichtbar bleiben.

  Die Arbeit leistet damit einen Beitrag zur systematischen Identifikation praxisrelevanter Problemfelder in der stationären Pflege sowie zur Ableitung fundierter Anforderungen an zukünftige technologische Unterstützungssysteme in diesem Pflegekontext.

  Literatur

  • Siehe Literatur für das Oberthema
  • Cameron-Mathiassen, J., Leiper, J., Simpson, J., & McDermott, E. (2022). What was care like for me? A systematic review of the experiences of young people living in residential care. Children and Youth Services Review, 138, 106524. https://doi.org/10.1016/j.childyouth.2022.106524
  • Giraldi, M., Mitchell, F., Porter, R. B., Reed, D., Jans, V., McIver, L., Manole, M., & McTier, A. (2022). Residential care as an alternative care option: A review of literature within a global context. Child & Family Social Work, 27(4), 825–837. https://doi.org/10.1111/cfs.12929
  • Smith, J., Bhandari, A., Yuksel, B., & Kocaballi, B. (2022). An Embodied Conversational Agent to Minimize the Effects of Social Isolation During Hospitalization. ACIS 2022 Proceedings. https://aisel.aisnet.org/acis2022/87
  • Huttunen, H.-L., Halonen, R., & Simon Klakegg. (2019). Proposal for Pervasive Elderly Care: A Case Study with Next of Kin. BLED 2019 Proceedings. https://aisel.aisnet.org/bled2019/59
  • Trainum, K., Tunis, R., Xie, B., & Hauser, E. (2023). Robots in Assisted Living Facilities: Scoping Review. JMIR Aging, 6, e42652. https://doi.org/10.2196/42652

• SOFTEC-BA-1-2, Sommersemester 2026, Betreuung: Jan Laufer , M. Sc.

  Zwischen Visite und Vitalwerten – KI und Robotik im Pflegealltag von Krankenhäusern

  Die Pflege in Krankenhäusern ist durch komplexe Arbeitsabläufe, hohe Arbeitsbelastung und eine Vielzahl medizinischer und organisatorischer Aufgaben geprägt. Pflegekräfte übernehmen eine zentrale Rolle in der Patientenversorgung, koordinieren Abläufe zwischen verschiedenen Berufsgruppen und stellen sicher, dass medizinische Maßnahmen zuverlässig umgesetzt werden.

  In den letzten Jahren wurden zunehmend technologische Lösungen entwickelt, die Pflegekräfte im Krankenhausalltag unterstützen sollen. Dazu gehören digitale Dokumentationssysteme, KI-basierte Entscheidungsunterstützung, intelligente Monitoring-Systeme oder logistische Robotersysteme. Dennoch ist bislang wenig darüber bekannt, welche Technologien tatsächlich im Pflegealltag eingesetzt werden und wie diese von Pflegekräften wahrgenommen werden.

  Diese Seminararbeit untersucht daher den Pflegealltag im Krankenhaus aus einer praxisnahen Perspektive.

  Forschungsfragen: 

  • Welche typischen Arbeitsabläufe und Aufgaben prägen den Alltag von Pflegekräften im Krankenhaus?
  • Welche Herausforderungen und Belastungen treten im Pflegealltag besonders häufig auf?
  • Welche Technologien werden derzeit im Krankenhausalltag eingesetzt?
  • Wie bewerten Pflegekräfte bestehende Technologien und digitale Systeme?
  • Welche Anforderungen bestehen an zukünftige Technologien zur Unterstützung der Pflege?

  Methodik:

  Die Arbeit folgt einem qualitativen Forschungsansatz und orientiert sich konzeptionell am echeloned Design Science Research (eDSR) Ansatz nach Tuunanen et al. (2024). Der eDSR-Ansatz strukturiert Design-Science-Forschung entlang mehrerer sogenannter Design Echelons, die unterschiedliche Ebenen der Problem- und Lösungsentwicklung adressieren. Während höhere Echelons typischerweise die Entwicklung, Implementierung und Evaluation konkreter Artefakte umfassen, konzentriert sich diese Seminararbeit auf die ersten beiden Echelons des Ansatzes. Die weiteren Echelons des eDSR-Ansatzes, die sich beispielsweise mit der konkreten Entwicklung, Implementierung oder Evaluation von Artefakten befassen, sind nicht Gegenstand dieser Seminararbeit und daher bewusst out of scope.

  Echelon 1 – Problem Exploration: In dieser Phase wird das Problemfeld systematisch untersucht, um ein fundiertes Verständnis der realen Praxisprobleme zu entwickeln. Ziel ist es, zentrale Herausforderungen, Arbeitsabläufe sowie bestehende Praktiken im Kontext der Krankenhauspflege zu identifizieren.

  Echelon 2 – Derivation of Design Knowledge: Auf Grundlage der identifizierten Problemfelder werden erste Designziele und Designanforderungen für potenzielle technologische Unterstützungssysteme abgeleitet.

  Zur Datensammlung werden qualitative Experteninterviews mit Pflegekräften aus dem jeweiligen Pflegekontext durchgeführt. Ziel der Interviews ist es, ein vertieftes Verständnis typischer Arbeitsabläufe, zentraler Herausforderungen sowie des aktuellen Einsatzes von Technologien im Pflegealltag zu gewinnen. Die Gestaltung und Durchführung der Interviews orientiert sich an etablierten Empfehlungen für qualitative Interviewforschung in der Informationssystemforschung (Myers & Newman, 2007).

  Zur Datenauswertung werden die Interviews transkribiert und anschließend mittels induktivem Open Coding analysiert. Ziel dieser Auswertung ist die systematische Identifikation zentraler Problemfelder sowie relevanter Anforderungen aus der Praxis. Die Analyse orientiert sich methodisch an der Grounded Theory Methodology, insbesondere am offenen, induktiven Kodieren qualitativer Daten (Birks et al., 2013).

  Erwarteter Beitrag:

  Die Arbeit liefert praxisnahe Einblicke in die Arbeitsrealität der Krankenhauspflege und identifiziert Potenziale sowie Grenzen für den Einsatz zukünftiger Technologien in diesem Bereich. Durch die Experteninterviews entsteht zunächst eine strukturierte und möglichst umfassende Übersicht über zentrale Probleme, Herausforderungen und Belastungen im Arbeitsalltag von Pflegekräften im Krankenhaus.

  Auf Basis dieser Problemübersicht sollen anschließend konkrete Gestaltungsansätze für zukünftige technologische Unterstützungssysteme abgeleitet werden. Dazu gehören insbesondere die Formulierung von Designzielen sowie Designanforderungen für mögliche Technologien, etwa im Bereich digitaler Dokumentation, intelligenter Monitoring-Systeme, KI-gestützter Entscheidungsunterstützung oder robotischer Logistiksysteme innerhalb von Krankenhäusern.

  Dabei wird ausdrücklich berücksichtigt, dass nicht alle identifizierten Probleme durch Technologie adressiert werden können. Ein wichtiger Teil der Analyse besteht daher darin zu unterscheiden, welche Herausforderungen prinzipiell technologisch unterstützbar sind und in welchen Bereichen menschliche Expertise, klinische Erfahrung und persönliche Interaktion mit Patientinnen und Patienten unverzichtbar bleiben.

  Die Arbeit leistet damit einen Beitrag zur systematischen Identifikation praxisrelevanter Problemfelder in der Krankenhauspflege sowie zur Ableitung fundierter Anforderungen an zukünftige technologische Unterstützungssysteme im klinischen Pflegekontext.

  Literatur

  • Siehe Literatur für das Oberthema
  • Smith, J., Bhandari, A., Yuksel, B., & Kocaballi, B. (2022). An Embodied Conversational Agent to Minimize the Effects of Social Isolation During Hospitalization. ACIS 2022 Proceedings. https://aisel.aisnet.org/acis2022/87
  • Wickramasinghe, N. (2019). A Mobile Nursing Solution. AMCIS 2019 Proceedings. https://aisel.aisnet.org/amcis2019/treo/treos/41

• SOFTEC-BA-1-3, Sommersemester 2026, Betreuung: Jan Laufer , M. Sc.

  Smart Home, Smart Care – Technologische Unterstützung in der ambulanten Pflege

  Die ambulante Pflege ermöglicht es vielen pflegebedürftigen Menschen, weiterhin in ihrem eigenen Zuhause zu leben. Pflegekräfte unterstützen dabei bei medizinischen Maßnahmen, alltäglichen Tätigkeiten sowie bei der Koordination mit Angehörigen und weiteren Gesundheitsdienstleistern.

  Die Arbeit in der ambulanten Pflege ist häufig durch Zeitdruck, lange Wegezeiten und eine hohe organisatorische Komplexität geprägt. Digitale Technologien, mobile Dokumentationssysteme, KI-basierte Assistenzlösungen oder intelligente Assistenzsysteme im häuslichen Umfeld werden daher zunehmend als mögliche Unterstützung diskutiert.

  Gleichzeitig stellt sich die Frage, welche Technologien tatsächlich im Alltag genutzt werden, welche Herausforderungen bestehen und welche Anforderungen Pflegekräfte an zukünftige technologische Lösungen haben.

  Forschungsfragen: 

  • Welche typischen Arbeitsabläufe prägen den Alltag von Pflegekräften in der ambulanten Pflege?
  • Welche organisatorischen und praktischen Herausforderungen entstehen im Pflegealltag?
  • Welche Technologien und digitalen Systeme werden aktuell eingesetzt?
  • Welche Anforderungen und Wünsche haben Pflegekräfte an zukünftige technologische Unterstützungssysteme?
  • Wo sehen Pflegekräfte Grenzen für den Einsatz von Technologie im Pflegekontext?

  Methodik:

  Die Arbeit folgt einem qualitativen Forschungsansatz und orientiert sich konzeptionell am echeloned Design Science Research (eDSR) Ansatz nach Tuunanen et al. (2024). Der eDSR-Ansatz strukturiert Design-Science-Forschung entlang mehrerer sogenannter Design Echelons, die unterschiedliche Ebenen der Problem- und Lösungsentwicklung adressieren. Während höhere Echelons typischerweise die Entwicklung, Implementierung und Evaluation konkreter Artefakte umfassen, konzentriert sich diese Seminararbeit auf die ersten beiden Echelons des Ansatzes. Die weiteren Echelons des eDSR-Ansatzes, die sich beispielsweise mit der konkreten Entwicklung, Implementierung oder Evaluation von Artefakten befassen, sind nicht Gegenstand dieser Seminararbeit und daher bewusst out of scope.

  Echelon 1 – Problem Exploration: In dieser Phase wird das Problemfeld systematisch untersucht, um ein fundiertes Verständnis der realen Praxisprobleme zu entwickeln. Ziel ist es, zentrale Herausforderungen, Arbeitsabläufe sowie bestehende Praktiken im Kontext ambulanter Pflege zu identifizieren.

  Echelon 2 – Derivation of Design Knowledge: Auf Grundlage der identifizierten Problemfelder werden erste Designziele und Designanforderungen für potenzielle technologische Unterstützungssysteme abgeleitet.

  Zur Datensammlung werden qualitative Experteninterviews mit Pflegekräften aus dem jeweiligen Pflegekontext durchgeführt. Ziel der Interviews ist es, ein vertieftes Verständnis typischer Arbeitsabläufe, zentraler Herausforderungen sowie des aktuellen Einsatzes von Technologien im Pflegealltag zu gewinnen. Die Gestaltung und Durchführung der Interviews orientiert sich an etablierten Empfehlungen für qualitative Interviewforschung in der Informationssystemforschung (Myers & Newman, 2007).

  Zur Datenauswertung werden die Interviews transkribiert und anschließend mittels induktivem Open Coding analysiert. Ziel dieser Auswertung ist die systematische Identifikation zentraler Problemfelder sowie relevanter Anforderungen aus der Praxis. Die Analyse orientiert sich methodisch an der Grounded Theory Methodology, insbesondere am offenen, induktiven Kodieren qualitativer Daten (Birks et al., 2013).

  Erwarteter Beitrag:

  Die Arbeit liefert praxisnahe Einblicke in die Arbeitsrealität der ambulanten Pflege und identifiziert Potenziale sowie Grenzen für den Einsatz zukünftiger Technologien in diesem Bereich. Durch die Experteninterviews entsteht zunächst eine strukturierte und möglichst umfassende Übersicht über zentrale Probleme, Herausforderungen und Belastungen im Arbeitsalltag von Pflegekräften in ambulanten Pflegediensten.

  Auf Basis dieser Problemübersicht sollen anschließend konkrete Gestaltungsansätze für zukünftige technologische Unterstützungssysteme abgeleitet werden. Dazu gehören insbesondere die Formulierung von Designzielen sowie Designanforderungen für mögliche Technologien, beispielsweise im Bereich mobiler Dokumentationssysteme, digitaler Planungs- und Koordinationstools, KI-gestützter Assistenzlösungen oder intelligenter Assistenzsysteme im häuslichen Umfeld von Pflegebedürftigen.

  Dabei wird ausdrücklich berücksichtigt, dass nicht alle identifizierten Probleme durch Technologie adressiert werden können. Ein wichtiger Teil der Analyse besteht daher darin zu unterscheiden, welche Herausforderungen prinzipiell technologisch unterstützbar sind und in welchen Bereichen persönliche Betreuung, Vertrauen sowie menschliche Interaktion eine zentrale Rolle spielen.

  Die Arbeit leistet damit einen Beitrag zur systematischen Identifikation praxisrelevanter Problemfelder in der ambulanten Pflege sowie zur Ableitung fundierter Anforderungen an zukünftige technologische Unterstützungssysteme im Kontext der ambulanten Pflege.

  Literatur

  • Siehe Literatur für das Oberthema
  • Muratore, A., & Ofe, H. (2025). DIGITAL MONITORING SERVICES FOR HOME CARE: INCENTIVISING ACTIVE PARTICIPATION OF ELDERLY PATIENTS THROUGH GAMIFIED SOLUTIONS. ECIS 2025 TREOs. https://aisel.aisnet.org/treos_ecis2025/14
  • Wickramasinghe, N. (2019). A Mobile Nursing Solution. AMCIS 2019 Proceedings. https://aisel.aisnet.org/amcis2019/treo/treos/41

In den letzten Jahren haben sich digitale Plattformen als zentrale Mechanismen der Wertschöpfung in unterschiedlichen Branchen etabliert. Unternehmen wie Apple, Amazon und Google zeigen, dass die strategische Orchestrierung von Wertschöpfungsprozessen über digitale Plattformen erhebliche Wettbewerbsvorteile ermöglicht. Ein zentraler Erfolgsfaktor plattformbasierter Ökosysteme ist die Integration komplementärer Angebote lose gekoppelter, unabhängiger Akteure. Plattformbetreiber fungieren als Orchestratoren, die die Interessen heterogener Stakeholder koordinieren müssen. Vor diesem Hintergrund gibt dieser Abschnitt einen Überblick über den aktuellen Forschungsstand zur Plattform-Governance. Die Beiträge dieses Themenblocks befassen sich mit der vertieften Analyse aufkommender Forschungsstränge sowie mit der empirischen Untersuchung von Governance-Mechanismen spezifischer Plattformökosysteme.

Liste der möglichen konkreten Themen:

• SOFTEC-BA-2-1, Sommersemester 2026, Betreuung: Robert Woroch , M. Sc.

  Zwischen Kreativität und Kontrolle: Governance in GenAI-Plattformökosystemen

  Generative künstliche Intelligenz (GenAI) nutzt tiefe generative Modelle, um auf Basis einfacher Nutzereingaben neuartige Inhalte in Domänen wie Text, Bilder, Video und Code zu erzeugen (Banh & Strobel, 2023). Im Gegensatz zu traditionellen KI-Systemen, die primär auf Vorhersage und Mustererkennung ausgerichtet sind, kann GenAI Kontext verstehen, aus Beispielen lernen und domänenübergreifend neue Inhalte generieren (Wessel et al., 2025).

  Das Aufkommen von GenAI stellt einen disruptiven Wandel für digitale Plattformen dar und verändert deren Funktionsweise sowie Wertschöpfung grundlegend. Durch die autonome Generierung neuer Ergebnisse ergeben sich weitreichende Implikationen für Plattformarchitektur, Governance und Stakeholder-Interaktionen. Insbesondere transformieren GenAI-Plattformen die Wertschöpfung durch Automatisierung, Demokratisierung der Partizipation, Hyperpersonalisierung und Mensch–KI-Kollaboration, wodurch sowohl Umfang als auch Komplexität steigen.

  Plattformbetreiber orchestrieren Ökosysteme, um deren Wertversprechen zu steigern (Kindermann et al., 2022). Dies erfolgt über Plattform-Governance-Mechanismen, verstanden als Aktivitäten zur Gestaltung der Ökosystemfunktion (Chen et al., 2022; Rietveld & Schilling, 2021). Im Gegensatz zu „Command-and-Control“-Ansätzen basieren diese auf „Connect-and-Coordinate“-Mechanismen zur Steuerung autonomer Akteure (Tilson et al., 2010).

  Im Kontext von GenAI müssen Boundary Resources und Anreizstrukturen angepasst werden, um menschliche Entwickler und agentische Komplementoren zu integrieren, etwa durch agentenorientierte Schnittstellen, Inter-Agenten-Protokolle und generative APIs sowie neue Erlösmodelle (Mayer et al., 2025).

  Zugleich entstehen spezifische Risiken wie Halluzinationen, Jailbreaking sowie Herausforderungen bei Datenqualität und sensiblen Informationen, die neue Governance-Mechanismen erfordern (Hein et al., 2020; Taeihagh, 2025). Darüber hinaus erhöhen Hyperpersonalisierung und autonome Agenten sowohl den Wertbeitrag als auch regulatorische Anforderungen, etwa in Bezug auf Datenschutz, Manipulation und bestehende Governance-Logiken (Feuerriegel et al., 2024; Wessel et al., 2025).

  Forschungsfrage:
  Welche Governance-Mechanismen implementieren Betreiber von GenAI-Plattformen zur Orchestrierung der Wertschöpfung in ihren Ökosystemen?

  Zielsetzung:
  Ziel dieser Arbeit ist die vollständige Entwicklung einer Taxonomie von Governance-Mechanismen in GenAI-Plattformökosystemen. Methodisch folgt die Arbeit dem Ansatz von Nickerson et al. (2013), erweitert durch Kundisch et al. (2022), und kombiniert konzeptionell-empirische sowie empirisch-konzeptionelle Iterationen.

  Die erste Iteration basiert auf einer vom Lehrstuhl bereitgestellten Übersicht von Governance-Mechanismen, die im Rahmen einer systematischen Literaturrecherche erstellt wurde. Darauf aufbauend werden zunächst fünf GenAI-Plattformen (z. B. OpenAI, Perplexity, Anthropic) analysiert, um Instanzen von Anreizmechanismen, Kontrollmechanismen und Boundary Ressourcen zu identifizieren.

  Anschließend werden mindestens zwei weitere Iterationen mit jeweils mindestens fünf zusätzlichen Plattformen durchgeführt, um die Taxonomie schrittweise weiterzuentwickeln. Für jede Iteration werden die Auswahlkriterien der Plattformen, die Erfüllung der Endbedingungen sowie die vorgenommenen Anpassungen der Taxonomie dokumentiert.

  Literatur

  • Banh, L., & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets, 33(1), 1–17. doi.org/10.1007/s12525-023-00680-1
  • Chen, L., Yi, J., Li, S., & Tong, T. W. (2022). Platform Governance Design in Platform Ecosystems: Implications for Complementors’ Multihoming Decision. Journal of Management, 48(3), 630–656.
  • Feuerriegel, S., Hartmann, J., Janiesch, C., & Zschech, P. (2024). Generative AI. Business & Information Systems Engineering, 66(1), 111–126. doi.org/10.1007/s12599-023-00834-7
  • Hein, A., Schreieck, M., Riasanow, T., Setzke, D. S., Wiesche, M., Böhm, M., & Krcmar, H. (2020). Digital platform ecosystems. Electronic Markets, 30(1), 87–98. doi.org/10.1007/s12525-019-00377-4
  • Kindermann, B., Salge, T. O., Wentzel, D., Flatten, T. C., & Antons, D. (2022). Dynamic capabilities for orchestrating digital innovation ecosystems: Conceptual integration and research opportunities. Information and Organization, 32(3).
  • Kundisch, D., Muntermann, J., Oberländer, A. M., Rau, D., Röglinger, M., Schoormann, T., & Szopinski, D. (2022). An Update for Taxonomy Designers. Business & Information Systems Engineering, 64(4), 421–439. doi.org/10.1007/s12599-021-00723-x
  • Mayer, A. S., Kostis, A., Strich, F., & Holmström, J. (2025). Shifting Dynamics: How Generative AI as a Boundary Resource Reshapes Digital Platform Governance. Journal of Management Information Systems, 42(2), 400–430.
  • Nickerson, R. C., Varshney, U., & Muntermann, J. (2013). A method for taxonomy development and its application in information systems. European Journal of Information Systems, 22(3), 336–359. doi.org/10.1057/ejis.2012.26
  • Rietveld, J., & Schilling, M. A. (2021). Platform Competition: A Systematic and Interdisciplinary Review of the Literature. Journal of Management, 47(6), 1528–1563.
  • Taeihagh, A. (2025). Governance of Generative AI. Policy and Society, 44(1), 1–22. doi.org/10.1093/polsoc/puaf001
  • Tilson, D., Lyytinen, K., & Sørensen, C. (2010). Research Commentary—Digital Infrastructures: The Missing IS Research Agenda. Information Systems Research, 21, 748–759.
  • Wessel, M., Adam, M., Benlian, A., Majchrzak, A., & Thies, F. (2025). Generative AI and its Transformative Value for Digital Platforms. Journal of Management Information Systems, 42(2), 346–369. doi.org/10.1080/07421222.2025.2487315

• SOFTEC-BA-2-2, Sommersemester 2026, Betreuung: Robert Woroch , M. Sc.

  Wenn Plattformen entscheiden: Algorithmische Governance im Fokus – Eine Forschungsagenda für digitale Ökosysteme

  Digitale Plattformökosysteme haben sich zu einer dominanten Organisationsform interorganisationaler Wertschöpfung entwickelt. Große Technologieunternehmen wie Alphabet, Meta, Apple, Microsoft und Amazon nutzen die Skalierbarkeit und Generativität ihrer Plattformen für Wachstum, während Start-ups erfolgreiche Plattformen in Bereichen wie künstliche Intelligenz (KI), E-Commerce und digitale Zahlungen etablieren.

  Digitale Plattformen sind Bündel digitaler Ressourcen, die wertschöpfende Interaktionen zwischen externen Akteuren als Produzenten und Konsumenten ermöglichen (Constantinides et al., 2018). Daraus entstehen komplexe Ökosysteme mit weitgehend autonomen Akteuren und heterogenen Zielen (Adner, 2017).

  Plattformbetreiber orchestrieren diese Ökosysteme, um das Wertversprechen zu steigern (Kindermann et al., 2022). Dies erfolgt über Plattform-Governance-Mechanismen (Rietveld & Schilling, 2021), also Aktivitäten zur Gestaltung der Ökosystemfunktion (Chen et al., 2022). Im Unterschied zu „Command-and-Control“-Ansätzen basiert dies auf „Connect-and-Coordinate“-Mechanismen zur Steuerung nicht hierarchisch kontrollierter Akteure (Tilson et al., 2010).

  Algorithmische Governance ist dabei zentral, da Plattformen zunehmend KI-gestützte Koordination und Entscheidungsfindung integrieren. Algorithmen fungieren als aktive Vermittler, die Daten interpretieren, Handlungen empfehlen und Entscheidungen treffen (Wessel et al., 2025). Plattformbetreiber müssen daher Systeme steuern, die Interaktionen prägen, und zugleich Fairness, Zuverlässigkeit und Rechenschaftspflicht gewährleisten.

  Die verstärkte Nutzung KI-basierter Bewertungen erhöht jedoch Bedenken hinsichtlich Fairness und Diskriminierung (Rosenblat & Stark, 2015; Wiener et al., 2023). Gleichzeitig schaffen Transparenz- und Erklärbarkeitsanforderungen neue Spannungsfelder, da zu viel Offenheit Überforderung oder Manipulation begünstigen kann (Zhang et al., 2022).

  Eine zentrale Herausforderung bleibt die algorithmische Opazität, da Einblicke in Bewertungs- und Entscheidungslogiken oft fehlen (Kellogg et al., 2020; Möhlmannn et al., 2023). Dies untergräbt Vertrauen und erschwert Fairnessbewertungen. Zudem reduziert Automatisierung menschliche Interaktion und kann Isolation fördern (Möhlmann et al., 2021; Wiener et al., 2023).

  Darüber hinaus nutzen Plattformen algorithmische Nudging-Mechanismen wie personalisierte Hinweise oder Gamification zur Verhaltenssteuerung. Diese sind effizient, bergen jedoch Manipulationsrisiken und können Machtasymmetrien verstärken, wenn sie nicht transparent reguliert werden (Benlian et al., 2022). Zugleich verschärft die zunehmende Nutzung algorithmischer Bewertungen die Herausforderungen in Bezug auf Fairness, Verzerrungen und Rechenschaftspflicht (Rosenblat & Stark, 2015; Wiener et al., 2023).

  Forschungsfrage:

  Welche zentralen Forschungsstränge lassen sich in der bestehenden Literatur zur algorithmischen Governance in digitalen Plattformökosystemen identifizieren, und welche offenen Forschungsfragen bestehen?

  Zielsetzung:

  Vor diesem Hintergrund zielt diese Arbeit darauf ab, eine Forschungsagenda zur algorithmischen Governance in digitalen Plattformökosystemen zu entwickeln. Hierzu werden zunächst zentrale Forschungsstränge der bestehenden Literatur identifiziert und systematisiert. Besonderes Augenmerk gilt dabei Herausforderungen in Transaktions- und Innovationsplattformen (Gawer, 2014; Hein et al., 2020) sowie unterschiedlichen Anwendungsdomänen.

  Darauf aufbauend wird für jeden Forschungsstrang eine spezifische Forschungsagenda entwickelt, die zentrale Herausforderungen herausarbeitet und vielversprechende zukünftige Forschungsrichtungen aufzeigt.

  Zur Abgrenzung des relevanten Literaturkorpus wird ein systematisches Literaturreview (SLR) gemäß Webster and Watson (2002) und vom Brocke et al. (2015). Die Analyse erfolgt mittels qualitativer Kodierung auf Basis etablierter methodischer Ansätze (Bandara et al., 2015; Wolfswinkel et al., 2013). Der Einsatz von Literaturverwaltungssoftware (z. B. Zotero oder Citavi) und qualitativer Analysesoftware (z. B. MAXQDA) ist obligatorisch.

  Literatur

  • Adner, R. (2017). Ecosystem as Structure. Journal of Management, 43(1), 39–58.
  • Bandara, W., Furtmueller, E., Gorbacheva, E., Miskon, S., & Beekhuyzen, J. (2015). Achieving Rigor in Literature Reviews: Insights from Qualitative Data Analysis and Tool-Support. Communications of the Association for Information Systems, 37.
  • Benlian, A., Wiener, M., Cram, W. A., Krasnova, H., Maedche, A., Möhlmann, M., Recker, J., & Remus, U. (2022). Algorithmic Management. Business & Information Systems Engineering, 64(6), 825–839.
  • Chen, L., Yi, J., Li, S., & Tong, T. W. (2022). Platform Governance Design in Platform Ecosystems: Implications for Complementors’ Multihoming Decision. Journal of Management, 48(3), 630–656.
  • Constantinides, P., Henfridsson, O., & Parker, G. G. (2018). Introduction—Platforms and Infrastructures in the Digital Age. Information Systems Research, 29(2), 381–400.
  • Gawer, A. (2014). Bridging differing perspectives on technological platforms: Toward an integrative framework. Research Policy, 43(7), 1239–1249. doi.org/10.1016/j.respol.2014.03.006
  • Hein, A., Schreieck, M., Riasanow, T., Setzke, D. S., Wiesche, M., Böhm, M., & Krcmar, H. (2020). Digital platform ecosystems. Electronic Markets, 30(1), 87–98. doi.org/10.1007/s12525-019-00377-4
  • Kellogg, K. C., Valentine, M. A., & Christin, A. (2020). Algorithms at Work: The New Contested Terrain of Control. Academy of Management Annals, 14(1), 366–410.
  • Kindermann, B., Salge, T. O., Wentzel, D., Flatten, T. C., & Antons, D. (2022). Dynamic capabilities for orchestrating digital innovation ecosystems: Conceptual integration and research opportunities. Information and Organization, 32(3).
  • Möhlmann, M., Zalmanson, L., Henfridsson, O., & Gregory, R. W. (2021). Algorithmic Management of Work on Online Labor Platforms: When Matching Meets Control. MIS Quarterly, 45(4), 1999–2022.
  • Möhlmannn, M., Salge, C. A. d. L., & Marabelli, M. (2023). Algorithm Sensemaking: How Platform Workers Make Sense of Algorithmic Management. Journal of the Association for Information Systems, 24(1), 35–64.
  • Rietveld, J., & Schilling, M. A. (2021). Platform Competition: A Systematic and Interdisciplinary Review of the Literature. Journal of Management, 47(6), 1528–1563.
  • Rosenblat, A., & Stark, L. (2015). Uber's Drivers: Information Asymmetries and Control in Dynamic Work. SSRN Electronic Journal.
  • Tilson, D., Lyytinen, K., & Sørensen, C. (2010). Research Commentary—Digital Infrastructures: The Missing IS Research Agenda. Information Systems Research, 21, 748–759.
  • vom Brocke, J., Simons, A., Riemer, K., Niehaves, B., Plattfaut, R., & Cleven, A. (2015). Standing on the Shoulders of Giants: Challenges and Recommendations of Literature Search in Information Systems Research. Communications of the Association for Information Systems, 37(1).
  • Webster, J., & Watson, R. T. (2002). Analyzing the Past to Prepare for the Future: Writing a Literature Review. MIS Quarterly, 26(2), xiii–xxiii.
  • Wessel, M., Adam, M., Benlian, A., Majchrzak, A., & Thies, F. (2025). Generative AI and its Transformative Value for Digital Platforms. Journal of Management Information Systems, 42(2), 346–369.
  • Wiener, M., Cram, W. A., & Benlian, A. (2023). Algorithmic control and gig workers: a legitimacy perspective of Uber drivers. European Journal of Information Systems, 32(3), 485–507.
  • Wolfswinkel, J. F., Furtmueller, E., & Wilderom, C. P. M. (2013). Using grounded theory as a method for rigorously reviewing literature. European Journal of Information Systems, 22(1), 45–55.
  • Zhang, A., Boltz, A., Wang, C. W., & Lee, M. K. (2022). Algorithmic Management Reimagined For Workers and By Workers: Centering Worker Well-Being in Gig Work. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems.

• SOFTEC-BA-2-3, Sommersemester 2026, Betreuung: Robert Woroch , M. Sc.

  Komplementoren im Fokus: Eine Evaluation von Wertenetzwerken generativer KI-Plattformen

  Generative künstliche Intelligenz (GenAI) nutzt tiefe generative Modelle, um auf Basis einfacher Nutzereingaben neuartige Inhalte in Domänen wie Text, Bilder, Video und Code zu erzeugen (Banh & Strobel, 2023). Im Gegensatz zu traditionellen KI-Systemen, die primär auf Vorhersage und Mustererkennung ausgerichtet sind, kann GenAI Kontext verstehen, aus Beispielen lernen und domänenübergreifend neue Inhalte generieren (Wessel et al., 2025).

  Digitale Plattformen lassen sich als Bündel digitaler Ressourcen verstehen, die wertschöpfende Interaktionen zwischen externen Akteuren auf der Angebots- und Nachfrageseite ermöglichen (Wessel et al., 2025). Auf dieser Grundlage entstehen komplexe Ökosysteme, die durch weitgehend autonome Akteure und heterogene Zielsetzungen gekennzeichnet sind (Adner, 2017). Im Kontext generativer KI besteht die zentrale Wertschöpfung in der Bereitstellung und Nutzung generativer Technologien zur Erstellung von Inhalten, zur kontextsensitiven Anpassung sowie zur natürlichen Interaktion. Diese Fähigkeiten fördern intelligente Automatisierung, Hyperpersonalisierung und kollaborative Innovation und begünstigen damit die Entstehung neuer Formen digitaler Dienstleistungen (Adner, 2017; Wessel et al., 2025).

  Aus organisationaler Perspektive lassen sich Plattformökosysteme als Meta-Organisationen begreifen, die durch einen geringen Grad an Formalisierung bei zugleich hoher Interdependenz der beteiligten Akteure gekennzeichnet sind (Kretschmer et al., 2022). Zu diesen Akteuren zählen komplementäre Anbieter, etwa App-Entwickler oder digitale Plattformen aus unterschiedlichen Domänen. Die Koordination dieser Akteure erfolgt mit dem Ziel, die Realisierung eines gemeinsamen Wertversprechens sicherzustellen (Adner, 2017; Hein et al., 2020).

  Zur Analyse solcher Systeme bieten generische Wertenetzwerke eine geeignete konzeptionelle Grundlage. Diese abstrahieren von konkreten Instanzen durch rollenbasierte Generalisierung und stellen wiederverwendbare Designartefakte dar, die als analytische Instrumente zur Untersuchung von Wertschöpfungsstrukturen dienen (Gregor & Hevner, 2013; Pousttchi & Gleiss, 2019; vom Brocke et al., 2020).

  Forschungsfrage:

  Wie lassen sich generative KI-Plattformökosysteme aus einer Wertenetzwerkperspektive konzeptualisieren?

  Zielsetzung:

  Ziel der Arbeit ist die Evaluation eines bereits entwickelten generischen Wertenetzwerks für GenAI-Plattformökosysteme, das auf Basis einer umfassenden empirischen Studie abgeleitet wurde. Hierzu wird zunächst ein strukturierter Evaluationsansatz entwickelt, der geeignete Kriterien und Methoden zur Bewertung des Artefakts definiert und sich an etablierten Frameworks der Design Science Research orientiert (Prat et al., 2015; Schoormann et al., 2024; Venable et al., 2016).

  Zur Datenerhebung werden qualitative Experteninterviews mit Akteuren durchgeführt, die GenAI-basierte Anwendungen entwickeln, betreiben oder durch komplementäre Angebote erweitern, darunter etwa App-Entwickler, Plattformbetreiber sowie Anbieter von Erweiterungslösungen. Ein vorgelagertes Zwischenziel besteht in der Identifikation geeigneter Experten, beispielsweise über Plattformen wie LinkedIn oder über Unternehmenswebsites. Ziel ist es, das Wertenetzwerk hinsichtlich seiner Rollen, Aktivitäten und Austauschbeziehungen sowie in Bezug auf Verständlichkeit, Vollständigkeit und wahrgenommene Nützlichkeit zu evaluieren.

  Die erhobenen Daten werden transkribiert und mittels kombinierter deduktiver und induktiver Codierung ausgewertet. Die Analyse orientiert sich methodisch an der Grounded Theory Methodology und dient der systematischen Identifikation zentraler Anforderungen sowie potenzieller Verbesserungspotenziale des Modells (Birks et al., 2013; Corbin & Strauss, 2015; Limpaecher & Ho, 2021; Saldaña, 2013).

  Die Arbeit leistet damit einen Beitrag zum besseren Verständnis der Wertkonfiguration und Interaktionsstrukturen in GenAI-Plattformökosystemen und unterstützt die Weiterentwicklung generischer Wertenetzwerke als analytische und gestaltungsorientierte Artefakte in der Plattformforschung.

  Literatur

  • Adner, R. (2017). Ecosystem as Structure. Journal of Management, 43(1), 39–58. doi.org/10.1177/0149206316678451
  • Banh, L., & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets, 33(1), 1–17. doi.org/10.1007/s12525-023-00680-1
  • Birks, D. F., Fernandez, W., Levina, N., & Nasirin, S. (2013). Grounded theory method in information systems research: its nature, diversity and opportunities. European Journal of Information Systems, 22(1), 1–8. doi.org/10.1057/ejis.2012.48
  • Corbin, J. M., & Strauss, A. L. (2015). Basics of qualitative research: Techniques and procedures for developing grounded theory (Fourth edition). Sage.
  • Gregor, S., & Hevner, A. R. (2013). Positioning and Presenting Design Science Research for Maximum Impact. MIS Quarterly, 37(2), 337–355. doi.org/10.25300/MISQ/2013/37.2.01
  • Hein, A., Schreieck, M., Riasanow, T., Setzke, D. S., Wiesche, M., Böhm, M., & Krcmar, H. (2020). Digital platform ecosystems. Electronic Markets, 30(1), 87–98. doi.org/10.1007/s12525-019-00377-4
  • Kretschmer, T., Leiponen, A., Schilling, M., & Vasudeva, G. (2022). Platform ecosystems as meta‐organizations: Implications for platform strategies. Strategic Management Journal, 43(3), 405–424. doi.org/10.1002/smj.3250
  • Limpaecher, A., & Ho, L. (2021, April 27). Deductive and Inductive Coding. Delve, 2021. delvetool.com/blog/deductiveinductive
  • Pousttchi, K., & Gleiss, A. (2019). Surrounded by middlemen - how multi-sided platforms change the insurance industry. Electronic Markets, 29(4), 609–629. doi.org/10.1007/s12525-019-00363-w
  • Prat, N., Comyn-Wattiau, I., & Akoka, J. (2015). A Taxonomy of Evaluation Methods for Information Systems Artifacts. Journal of Management Information Systems, 32(3), 229–267. doi.org/10.1080/07421222.2015.1099390
  • Saldaña, J. (2013). The coding manual for qualitative researchers (2. ed.). SAGE Publ.
  • Schoormann, T., Möller, F., Chandra Kruse, L., & Otto, B. (2024). BAUSTEIN —A design tool for configuring and representing design research. Information Systems Journal, 34(6), 1871–1901. doi.org/10.1111/isj.12516
  • Venable, J., Pries-Heje, J., & Baskerville, R. (2016). FEDS: a Framework for Evaluation in Design Science Research. European Journal of Information Systems, 25(1), 77–89. doi.org/10.1057/ejis.2014.36
  • vom Brocke, J., Winter, R., Hevner, A., & Maedche, A. (2020). Special Issue Editorial –Accumulation and Evolution of Design Knowledge in Design Science Research: A Journey Through Time and Space. Journal of the Association for Information Systems, 21(3), 520–544. doi.org/10.17705/1jais.00611
  • Wessel, M., Adam, M., Benlian, A., Majchrzak, A., & Thies, F. (2025). Generative AI and its Transformative Value for Digital Platforms. Journal of Management Information Systems, 42(2), 346–369.

• SOFTEC-BA-2-4, Sommersemester 2026, Betreuung: Robert Woroch , M. Sc.

  Die Rolle von Service-Providern in generativen KI-Ökosystemen: Eine Evaluation digitaler Wertenetzwerke

  Generative künstliche Intelligenz (GenAI) nutzt tiefe generative Modelle, um auf Basis einfacher Nutzereingaben neuartige Inhalte in Domänen wie Text, Bilder, Video und Code zu erzeugen (Banh & Strobel, 2023). Im Gegensatz zu traditionellen KI-Systemen, die primär auf Vorhersage und Mustererkennung ausgerichtet sind, kann GenAI Kontext verstehen, aus Beispielen lernen und domänenübergreifend neue Inhalte generieren (Wessel et al., 2025).

  Digitale Plattformen lassen sich als Bündel digitaler Ressourcen verstehen, die wertschöpfende Interaktionen zwischen externen Akteuren auf der Angebots- und Nachfrageseite ermöglichen (Wessel et al., 2025). Auf dieser Grundlage entstehen komplexe Ökosysteme, die durch weitgehend autonome Akteure und heterogene Zielsetzungen gekennzeichnet sind (Adner, 2017). Im Kontext generativer KI besteht die zentrale Wertschöpfung in der Bereitstellung und Nutzung generativer Technologien zur Erstellung von Inhalten, zur kontextsensitiven Anpassung sowie zur natürlichen Interaktion. Diese Fähigkeiten fördern intelligente Automatisierung, Hyperpersonalisierung und kollaborative Innovation und begünstigen damit die Entstehung neuer Formen digitaler Dienstleistungen (Adner, 2017; Wessel et al., 2025).

  Aus organisationaler Perspektive lassen sich Plattformökosysteme als Meta-Organisationen begreifen, die durch einen geringen Grad an Formalisierung bei zugleich hoher Interdependenz der beteiligten Akteure gekennzeichnet sind (Kretschmer et al., 2022). Zu diesen Akteuren zählen unter anderem Serviceanbieter, etwa im Bereich von Cloud-Rechenzentren, Finetuning oder Benchmarking-Lösungen. Die Koordination dieser Akteure erfolgt mit dem Ziel, die Realisierung eines gemeinsamen Wertversprechens sicherzustellen (Adner, 2017; Hein et al., 2020).

  Zur Analyse solcher Systeme bieten generische Wertenetzwerke eine geeignete konzeptionelle Grundlage. Diese abstrahieren von konkreten Instanzen durch rollenbasierte Generalisierung und stellen wiederverwendbare Designartefakte dar, die als analytische Instrumente zur Untersuchung von Wertschöpfungsstrukturen dienen (Gregor & Hevner, 2013; Pousttchi & Gleiss, 2019; vom Brocke et al., 2020).

  Forschungsfrage:

  Wie lassen sich generative KI-Plattformökosysteme aus einer Wertenetzwerkperspektive konzeptualisieren?

  Zielsetzung:

  Ziel der Arbeit ist die Evaluation eines bereits entwickelten generischen Wertenetzwerks für GenAI-Plattformökosysteme, das auf Basis einer umfassenden empirischen Studie abgeleitet wurde. Hierzu wird zunächst ein strukturierter Evaluationsansatz entwickelt, der geeignete Kriterien und Methoden zur Bewertung des Artefakts definiert und sich an etablierten Frameworks der Design Science Research orientiert (Prat et al., 2015; Schoormann et al., 2024; Venable et al., 2016).

  Zur Datenerhebung werden qualitative Experteninterviews mit Akteuren durchgeführt, die Rechenzentren für GenAI-basierte Anwendungen betreiben, Finetuning- oder Benchmarking-Lösungen anbieten sowie entsprechende Anwendungen entwickeln, betreiben oder durch komplementäre Angebote erweitern, darunter etwa App-Entwickler, Plattformbetreiber und Anbieter von Erweiterungslösungen. Ein vorgelagertes Zwischenziel besteht in der Identifikation geeigneter Experten, beispielsweise über Plattformen wie LinkedIn oder über Unternehmenswebsites. Ziel ist es, das Wertenetzwerk hinsichtlich seiner Rollen, Aktivitäten und Austauschbeziehungen sowie in Bezug auf Verständlichkeit, Vollständigkeit und wahrgenommene Nützlichkeit zu evaluieren.

  Die erhobenen Daten werden transkribiert und mittels kombinierter deduktiver und induktiver Codierung ausgewertet. Die Analyse orientiert sich methodisch an der Grounded Theory Methodology und dient der systematischen Identifikation zentraler Anforderungen sowie potenzieller Verbesserungspotenziale des Modells (Birks et al., 2013; Corbin & Strauss, 2015; Limpaecher & Ho, 2021; Saldaña, 2013).

  Die Arbeit leistet damit einen Beitrag zum besseren Verständnis der Wertkonfiguration und Interaktionsstrukturen in GenAI-Plattformökosystemen und unterstützt die Weiterentwicklung generischer Wertenetzwerke als analytische und gestaltungsorientierte Artefakte in der Plattformforschung.

  Literatur

  • Adner, R. (2017). Ecosystem as Structure. Journal of Management, 43(1), 39–58. doi.org/10.1177/0149206316678451
  • Banh, L., & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets, 33(1), 1–17. doi.org/10.1007/s12525-023-00680-1
  • Birks, D. F., Fernandez, W., Levina, N., & Nasirin, S. (2013). Grounded theory method in information systems research: its nature, diversity and opportunities. European Journal of Information Systems, 22(1), 1–8. doi.org/10.1057/ejis.2012.48
  • Corbin, J. M., & Strauss, A. L. (2015). Basics of qualitative research: Techniques and procedures for developing grounded theory (Fourth edition). Sage.
  • Gregor, S., & Hevner, A. R. (2013). Positioning and Presenting Design Science Research for Maximum Impact. MIS Quarterly, 37(2), 337–355. doi.org/10.25300/MISQ/2013/37.2.01
  • Hein, A., Schreieck, M., Riasanow, T., Setzke, D. S., Wiesche, M., Böhm, M., & Krcmar, H. (2020). Digital platform ecosystems. Electronic Markets, 30(1), 87–98. doi.org/10.1007/s12525-019-00377-4
  • Kretschmer, T., Leiponen, A., Schilling, M., & Vasudeva, G. (2022). Platform ecosystems as meta‐organizations: Implications for platform strategies. Strategic Management Journal, 43(3), 405–424. doi.org/10.1002/smj.3250
  • Limpaecher, A., & Ho, L. (2021, April 27). Deductive and Inductive Coding. Delve, 2021. delvetool.com/blog/deductiveinductive
  • Pousttchi, K., & Gleiss, A. (2019). Surrounded by middlemen - how multi-sided platforms change the insurance industry. Electronic Markets, 29(4), 609–629. doi.org/10.1007/s12525-019-00363-w
  • Prat, N., Comyn-Wattiau, I., & Akoka, J. (2015). A Taxonomy of Evaluation Methods for Information Systems Artifacts. Journal of Management Information Systems, 32(3), 229–267. doi.org/10.1080/07421222.2015.1099390
  • Saldaña, J. (2013). The coding manual for qualitative researchers (2. ed.). SAGE Publ.
  • Schoormann, T., Möller, F., Chandra Kruse, L., & Otto, B. (2024). BAUSTEIN —A design tool for configuring and representing design research. Information Systems Journal, 34(6), 1871–1901. doi.org/10.1111/isj.12516
  • Venable, J., Pries-Heje, J., & Baskerville, R. (2016). FEDS: a Framework for Evaluation in Design Science Research. European Journal of Information Systems, 25(1), 77–89. doi.org/10.1057/ejis.2014.36
  • vom Brocke, J., Winter, R., Hevner, A., & Maedche, A. (2020). Special Issue Editorial –Accumulation and Evolution of Design Knowledge in Design Science Research: A Journey Through Time and Space. Journal of the Association for Information Systems, 21(3), 520–544. doi.org/10.17705/1jais.00611
  • Wessel, M., Adam, M., Benlian, A., Majchrzak, A., & Thies, F. (2025). Generative AI and its Transformative Value for Digital Platforms. Journal of Management Information Systems, 42(2), 346–369.

Jüngste Fortschritte in der künstlichen Intelligenz, insbesondere durch Large Language Models (LLMs) und generative Agenten, erweitern die kognitiven Fähigkeiten von Informationssystemen (IS) grundlegend und verschieben damit das traditionelle Verständnis der Mensch-Maschine-Beziehung. Während IS bislang primär als reaktive Werkzeuge fungierten, die menschliche Entscheidungen unterstützen, sind moderne IS zunehmend in der Lage, eigenständig Ziele zu verfolgen, in unstrukturierten Umgebungen zu handeln und Aufgaben proaktiv an andere Akteure zu delegieren – ein Phänomen, das in der Forschung als Agentic IS bezeichnet wird. Agency meint dabei die Fähigkeit eines Systems, Rechte und Verantwortlichkeiten für Entscheidungen und Handlungen selbstständig zu übernehmen. Dieser Paradigmenwechsel eröffnet neuartige Formen der Zusammenarbeit zwischen Mensch und IS – etwa in kollaborativen oder hybriden Mensch-IS-Teams –, wirft jedoch zugleich grundlegende Fragen hinsichtlich Delegation, Vertrauen, menschlicher Kontrolle und rechtlicher Verantwortlichkeit auf. Wie aktuelle Forschung zeigt, sind sowohl die effektive Gestaltung von Delegationsprozessen als auch die Sicherstellung menschlicher Handlungsfähigkeit in agentischen Systemen bislang unzureichend verstanden. In diesem Themenkomplex sollen daher die Auswirkungen agentischer Informationssysteme auf Individuen, Organisationen und gesellschaftliche Strukturen systematisch untersucht werden.

Literatur

• Baird, A. & Maruping, L. M. (2021). The Next Generation of Research on IS Use: A Theoretical Framework of Delegation to and from Agentic IS Artifacts. MIS Quarterly, 45(1), 315–341. https://doi.org/10.25300/misq/2021/15882
• Fechner, P., Lämmermann, L., Lockl, J., Röglinger, M. & Urbach, N. (2025). F Toward Triadic Delegation: How Agentic IS Artifacts Affect the Patient-Doctor Relationship in Healthcare. Journal Of The Association For Information Systems, 26(6), 1703–1736. doi.org/10.17705/1jais.00954
• Holldack, F., Banh, L. & Strobel, G. (2026). Agentic information systems. Electronic Markets, 36(1). doi.org/10.1007/s12525-025-00861-0
• Kuss, P. & Meske, C. (2025). From Entity to Relation? Agency in the Era of Artificial Intelligence. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5183049
• Leonardi, P. M. (2025). Homo agenticus in the age of agentic AI: Agency loops, power displacement, and the circulation of responsibility. Information and Organization, 35(3), 100582. https:// doi.org/10.1016/j.infoandorg.2025.100582
• Schuetz, S. & Venkatesh, V. (2020). Research Perspectives: The Rise of Human Machines: How Cognitive Computing Systems Challenge Assumptions of User-System Interaction. Journal Of The Association For Information Systems, 460–482. doi.org/10.17705/1jais.00608
• Stelmaszak, M., Möhlmann, M., & Sørensen, C. (2025). When Algorithms Delegate to Humans: Exploring Human-Algorithm Interaction at Uber. MISQ, 49(1), 305–330. doi.org/10.25300/MISQ/2024/17911
• Strunk, J., Banh, L., Nissen, A., Strobel, G., & Smolnik, S. (2024). To Delegate or Not to Delegate? Factors Influencing Human-Agentic IS Interaction. ICIS 2024 Proceedings.

Liste der möglichen konkreten Themen:

• SOFTEC-BA-3-1, Sommersemester 2026, Betreuung: Florian Holldack , M. Sc.

  Wenn Systeme mitdenken: Wie Agentic IS das organisationale Wissensmanagement transformieren

  Organisationales Wissensmanagement basiert seit Jahrzehnten auf der grundlegenden Unterscheidung zwischen explizitem Wissen (kodifizierbares, speicherbarem Wissen) und implizitem (tacit) Wissen, das in menschlicher Erfahrung, Intuition und Praxis verankert ist. Traditionelle IS konnten ersteres effizient verwalten, stießen bei letzterem jedoch strukturell an ihre Grenzen. Mit dem Aufkommen agentischer Informationssysteme verschwimmt diese Grenze zunehmend: Agentic IS sind in der Lage, unstrukturierte Situationen zu navigieren, kontextabhängige Schlussfolgerungen zu ziehen und aktiv an wissensintensiven Prozessen teilzunehmen – Fähigkeiten, die bislang als Domäne menschlicher Expertise galten. Dies wirft grundlegende Fragen für das Wissensmanagement in Organisationen auf: Wie verändern agentische Systeme die Entstehung, Weitergabe und Anwendung von Wissen? Welche Rolle spielen sie bei der Externalisierung tacit knowledge, und welche Risiken entstehen, wenn organisationales Wissen zunehmend in IS-Artefakten statt in Menschen verankert ist? Obwohl erste Studien auf disruptive Effekte für wissensintensive Arbeit hindeuten, fehlt bislang eine systematische Aufarbeitung der Literatur zu den Auswirkungen agentischer IS auf organisationales Wissensmanagement.

  Ziel dieser Seminararbeit ist es daher, im Rahmen einer systematischen Literaturrecherche (SLR) relevante wissenschaftliche Arbeiten zu analysieren, die sich mit dem Einfluss agentischer IS auf Wissensmanagementprozesse in Organisationen befassen. Auf dieser Grundlage sollen zentrale Veränderungsmuster herausgearbeitet und ein konzeptioneller Rahmen entwickelt werden, der beschreibt, wie Agentic IS bestehende Wissensmanagementpraktiken ergänzen, transformieren oder ersetzen. Die Ergebnisse sollen sowohl theoretische Implikationen für die IS-Forschung liefern als auch praktische Orientierung für Organisationen bieten, die agentische Systeme in wissensintensive Prozesse integrieren möchten.

  Literatur

  • Alavi, M., Leidner, D. E. & Mousavi, R. (2024). Knowledge Management Perspective of Generative Artificial Intelligence. Journal Of The Association For Information Systems, 25(1), 1–12. doi.org/10.17705/1jais.00859
  • Heimburg, V., Yoo, Y. & Wiesche, M. (2025). Temporality of Organizational Knowledge in Generative AI Systems. ICIS 2025 Proceedings.
  • Holldack, F., Banh, L. & Strobel, G. (2026). Agentic information systems. Electronic Markets, 36(1). https://doi.org/10.1007/s12525-025-00861-0
  • Jussupow, E., Spohrer, K., Heinzl, A., & Gawlitza, J. (2021). Augmenting medical diagnosis decisions? An investigation into physicians’ decision-making process with artificial intelligence. Information Systems Research, 32(3), 713–735. doi.org 10.1287/isre.2020.0980
  • Sabherwal, R. & Grover, V. (2024). The Societal Impacts of Generative Artificial Intelligence: A Balanced Perspective. Journal Of The Association For Information Systems, 25(1), 13–22. doi.org/10.17705/1jais.00860
  • Wang, B. Y., Boell, S. & Sun, Y. (2025). Navigating the Complexities of Organizational Knowledge Management in the Age of Generative AI. ICIS 2025 Proceedings.
  • Wang, J. & Legner, C. (2025). Uncovering Untapped Organizational Knowledge in Unstructured Data: GenAI and the Reconfiguration of Data Management. ICIS 2025 Proceedings.

• SOFTEC-BA-3-2, Sommersemester 2026, Betreuung: Florian Holldack , M. Sc.

  Zwischen Autonomie und Kontrolle: Die Spannung zwischen wachsender IS-Agency und dem Bedarf menschlicher Steuerung

  Mit zunehmender Agency agentischer Informationssysteme verschärft sich ein fundamentales Spannungsfeld: Je autonomer ein IS handelt, desto schwieriger wird es für Menschen, effektive Kontrolle auszuüben, ohne dabei die Effizienzgewinne zu untergraben, die Autonomie erst ermöglicht. Regulatorische Rahmenbedingungen wie der EU AI Act schreiben für risikoreiche Anwendungen explizit menschliche Aufsicht vor, während die Forschung zeigt, dass übermäßige Kontrollmechanismen sowohl die Leistungsfähigkeit agentischer Systeme einschränken als auch zu Skill Erosion auf Seiten menschlicher Akteure führen können. Gleichzeitig sind unzureichende Kontrollmechanismen mit erheblichen Risiken verbunden – von Accountability-Lücken über Vertrauensverlust bis hin zu unkontrollierbaren Systementscheidungen in kritischen Kontexten. Obwohl Delegation, Vertrauen und Kontrolle zunehmend als zentrale Herausforderungen im Agentic IS-Paradigma anerkannt werden, fehlt bislang eine systematische Synthese darüber, wie diese Spannung in der wissenschaftlichen Literatur konzeptualisiert wird und welche Gestaltungs- sowie Governance-Prinzipien existieren, um ihr zu begegnen.

  Ziel dieser Seminararbeit ist es daher, im Rahmen einer systematischen Literaturrecherche (SLR) wissenschaftliche Arbeiten zu analysieren, die sich mit dem Spannungsverhältnis zwischen IS-Agency und menschlicher Steuerung befassen. Dabei sollen sowohl technische als auch organisationale und regulatorische Lösungsansätze – etwa Abstufungen menschlicher Aufsicht (Human-in-the-loop vs. Human-on-the-loop) oder vertrauensfördernde Systemgestaltung – systematisch aufgearbeitet werden. Die Ergebnisse sollen ein vertieftes Verständnis davon ermöglichen, unter welchen Bedingungen welche Kontrollmechanismen geeignet sind, und damit eine wissenschaftliche Grundlage für die verantwortungsvolle Gestaltung und Governance agentischer Systeme schaffen.

  Literatur

  • Diebel, C., Goutier, M., Adam, M., & Benlian, A. (2025). When AI-based agents are proactive: Implications for competence and system satisfaction in human–AI collaboration. Business & Information Systems Engineering. doi.org/10.1007/ s12599-024-00918-y
  • Fechner, P., Lämmermann, L., Lockl, J., Röglinger, M. & Urbach, N. (2025). F Toward Triadic Delegation: How Agentic IS Artifacts Affect the Patient-Doctor Relationship in Healthcare. Journal Of The Association For Information Systems, 26(6), 1703–1736. https://doi.org/10.17705/1jais.00954
  • Fügener, A., Grahl, J., Gupta, A., & Ketter, W. (2021). Cognitive challenges in human–artificial intelligence collaboration: Investigating the path toward productive delegation. Information Systems Research, 33(2), 678–696. doi.org/10. 1287/isre.2021.1079
  • Holldack, F., Banh, L. & Strobel, G. (2026). Agentic information systems. Electronic Markets, 36(1). https://doi.org/10.1007/s12525-025-00861-0
  • Höhener, Daria (2026). The Evolution of AI Compliance Assistance from Reactive Support to Co-Agency. MIS Quarterly Executive: Vol. 25: Iss. 1, Article 4.
  • Mihale-Wilson, C. A. (2025). From Whether to How Much: A Multi-tiered Perspective on Delegation to Agentic IS. ICIS 2025 Proceedings.
  • Recker, J., Chatterjee, S., Sundermeier, J. & Tarafdar, M. (2025). Digital Responsibility: Current Perspectives and Future Directions. Journal Of The Association For Information Systems, 26(5), 1222–1238. doi.org/10.17705/1jais.00966
  • Wissuchek, C. & Zschech, P. (2025). Challenges in Managing the Relationship Between Agentic AI Systems and Humans in Organizations. In Lecture notes in business information processing (S. 3–17). doi.org/10.1007/978-3-031-94193-1_1

Die rasante Verbreitung generativer KI verändert die Hochschullehre grundlegend und eröffnet neue Möglichkeiten für personalisierte, adaptive Lernunterstützung jenseits standardisierter Lernangebote. Large Language Models (LLMs) ermöglichen es, auf Basis von Lehrmaterialien fachspezifische Lernassistenten zu entwickeln, die Studierende individuell durch den Stoff führen, Wissensstand evaluieren und gezielt Feedback geben. Trotz dieses Potenzials mangelt es bestehenden GenAI-Tools wie ChatGPT an fachlicher Fundierung, didaktischer Verankerung und Integration in die Hochschullernumgebung. Forschung zeigt zudem, dass gut gestaltete GenAI-Lernassistenten selbstgesteuertes Lernen aktiv fördern können – etwa durch adaptive Inhalte, kollaborative Dialogformate und konstruktives Feedback –, ohne kritisches Denken zu ersetzen. Gleichzeitig stellen Fragen der Datenschutzkonformität, der verantwortungsvollen Nutzung und der AI Literacy zentrale Herausforderungen für eine nachhaltige Integration in die Hochschullehre dar.

Literatur

• Adam, M., Bauer, K., Jussupow, E. et al. Generating Tomorrow’s Me: How Collaborating with Generative AI Changes Humans. Bus Inf Syst Eng67, 583–594 (2025). doi.org/10.1007/s12599-025-00961-3
• Banh, L. & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets 33(63).
• Van Slyke, C., Johnson, R. D., & Sarabadani, J. (2023). Generative Artificial Intelligence in Information Systems Education: Challenges, Consequences, and Responses. Communications of the Association for Information Systems, 53, 1-21. https://doi.org/10.17705/1CAIS.05301

Liste der möglichen konkreten Themen:

• SOFTEC-BA-4-1, Sommersemester 2026, Betreuung: Leonardo Banh , M. Sc.

  Adaptives Lernen mit Generativer KI: Stand der Forschung und Gestaltungsempfehlungen für KI-Lernassistenten

  Mit dem Aufkommen generativer KI entstehen neue Möglichkeiten für personalisierte Lernunterstützung in der Hochschulbildung. Während klassische Lernmanagementsysteme standardisierte Inhalte bereitstellen, können LLM-basierte Assistenten adaptiv auf individuelle Lernstile, Vorkenntnisse und Lerntempi eingehen und Studierende aktiv durch den Stoff führen. Trotz dieses Potenzials mangelt es bestehenden GenAI-Tools wie ChatGPT an fachlicher Fundierung und didaktischer Verankerung – Aspekte, die für einen nachhaltigen Lernerfolg jedoch entscheidend sind. Obwohl erste Studien zeigen, dass gut gestaltete KI-Lernassistenten selbstgesteuertes Lernen fördern können, fehlt bisher eine konsolidierte Übersicht darüber, welche Gestaltungsprinzipien für KI-Lernassistenten in der Hochschullehre maßgeblich sind.

  Ziel dieser Seminararbeit ist es daher, im Rahmen einer systematischen Literaturrecherche (SLR) wissenschaftliche Arbeiten zu analysieren, die sich mit dem Einsatz generativer KI zur Lernunterstützung im Hochschulkontext befassen. Auf dieser Grundlage sollen zentrale didaktische Prinzipien und Gestaltungsmerkmale erfolgreicher KI-Lernassistenten herausgearbeitet werden. Die Ergebnisse sollen ein vertieftes Verständnis davon ermöglichen, wie GenAI-basierte Lernsysteme didaktisch fundiert und adaptiv gestaltet werden können, und damit eine wissenschaftliche Grundlage für die Entwicklung solcher Systeme schaffen.

  Literatur

  • Banh, L. & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets, 33(63). DOI: 10.1007/s12525-023-00680-1
  • Knowles, M. S. (1975). Self-directed learning: A guide for learners and teachers. Association Press.
  • Romero, M., Reyes, J., & Kostakos, P. (2024). Generative Artificial Intelligence in Higher Education. In: Creative Applications of Artificial Intelligence in Education. Springer, Cham.
  • Shah, C. S., Mathur, S., & Vishnoi, S. K. (2024). Is ChatGPT Enhancing Youth's Learning, Engagement and Satisfaction? Journal of Computer Information Systems, 1–16.
  • vom Brocke, J. et al. (2009). Reconstructing the giant: On the importance of rigour in documenting the literature search process. ECIS 2009 Proceedings.
  • Webster, J. & Watson, R. T. (2002). Analyzing the Past to Prepare for the Future: Writing a Literature Review. MIS Quarterly, 26(2), xiii–xxiii.

• SOFTEC-BA-4-2, Sommersemester 2026, Betreuung: Leonardo Banh , M. Sc.

  Datenschutzkonforme KI in der Hochschullehre: Anforderungen, Lösungsansätze und Implikationen für den Einsatz lokaler Sprachmodelle

  Der Einsatz generativer KI-Systeme in Bildungseinrichtungen wirft grundlegende datenschutzrechtliche Fragen auf. Insbesondere die DSGVO und der EU AI Act stellen spezifische Anforderungen an die Verarbeitung personenbezogener Daten von Studierenden, denen kommerzielle Anbieter wie ChatGPT in der Regel nicht ohne Weiteres gerecht werden. Lokale KI-Instanzen auf Basis von Retrieval-Augmented Generation (RAG) gelten dabei als vielversprechende Alternative, da sie nicht nur Datenschutzanforderungen erfüllen, sondern auch eine fachspezifische Ausrichtung auf geprüften Lehrmaterialien ermöglichen. Obwohl RAG-basierte Ansätze in der Forschung zunehmend Beachtung finden, sind die konkreten Anforderungen und Implikationen für den Hochschulbetrieb bislang wenig systematisch untersucht.

  Ziel dieser Seminararbeit ist es daher, auf Basis einer kombinierten Analyse rechtlicher Rahmenbedingungen und einer systematischen Literaturrecherche herauszuarbeiten, welche datenschutzrechtlichen Anforderungen beim KI-Einsatz in der Hochschullehre zu beachten sind und welche technischen sowie organisatorischen Lösungsansätze existieren. Optional können die Erkenntnisse durch Experteninterviews mit Hochschul-IT-Verantwortlichen ergänzt werden. Die Ergebnisse sollen konkrete Handlungsempfehlungen für Hochschulen liefern, die eigene datenschutzkonforme KI-Systeme entwickeln und betreiben möchten.

  Literatur

  • Banh, L. & Strobel, G. (2023). Generative artificial intelligence. Electronic Markets, 33(63). DOI: 10.1007/s12525-023-00680-1
  • Klesel, M. & Wittmann, H. F. (2025). Retrieval-Augmented Generation (RAG). Business & Information Systems Engineering, 67, 551–561. DOI: 10.1007/s12599-025-00945-3
  • Lewis, P. et al. (2020). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Advances in Neural Information Processing Systems, 33, 9459–9474.
  • Strobel, G., Banh, L., Möller, F., & Schoormann, T. (2024). Exploring Generative Artificial Intelligence: A Taxonomy and Types. Proceedings of the 57th HICSS.
  • vom Brocke, J. et al. (2009). Reconstructing the giant: On the importance of rigour in documenting the literature search process. ECIS 2009 Proceedings.
  • Webster, J. & Watson, R. T. (2002). Analyzing the Past to Prepare for the Future: Writing a Literature Review. MIS Quarterly, 26(2), xiii–xxiii.

The increasing digitization of organizations and services has led to a rapid growth in the availability of data, making data a central resource for innovation and value creation. Data is no longer viewed only as an input for decision-making but has become a key asset whose value emerges through its combination with other data types, technologies, and organizational capabilities leading to more data-driven decision making. In this context, the concept of data complementarity suggests that the value of data increases when different datasets can be integrated, reused, or jointly analyzed, often leading to not only higher informational value but also to the creation of entirely novel forms of data. 

Healthcare represents a particularly suitable context for studying such mechanisms, as it is characterized by a highly data-intensive yet fragmented ecosystem involving healthcare providers, patients, insurers, digital platforms, and medical devices. Traditionally, healthcare relied mainly on structured clinical and administrative data, but through digital transformation and regulatory oversight, a wide range of new data forms have been introduced. When these heterogeneous data sources are combined, they hold the potential in generating new datasets that did not exist before. From an IS perspective, such processes can be understood as novel value creation through complementing data, where the interaction between different data types, infrastructures, and analytics capabilities enables new services, knowledge, and digital innovations. In healthcare, these complementarities may lead to new forms of personalized medicine, improved diagnostics, and digital health platforms. At the same time, the integration of diverse data sources raises important challenges regarding interoperability, governance, data quality, cross-domain integration, and value appropriation.

Literatur

• Jacobides, M. G., Cennamo, C., & Gawer, A. (2024). Externalities and complementarities in platforms and ecosystems: From structural solutions to endogenous failures. Research Policy, 53(1), 104906.
• Kilgus, T., Patecka, A., Schurig, T., Kari, A., Gubser, R., Gersch, M., ... & Fürstenau, D. (2024). Creating value from the secondary use of health data: International examples, best practices, and opportunities to scale. Communications of the Association for Information Systems, 55(1), 507-534.
• Konopik, J. (2023). The impact of digital platforms and ecosystems in healthcare on value creation—a integrative review and research agenda. IEEE access, 11, 135811-135819.
• Nienstedt, J., & Trenz, M. (2025). From sharing to profiting: Exploring the interplay between value creation and strategic appropriation in data ecosystems. Electronic Markets, 35(1), 1-24.
• Ritala, P., & Karhu, K. (2023). Capturing value from data complementarities: A multi-level framework. In Research handbook on digital strategy (pp. 273-288). Edward Elgar Publishing.
• Yoo, Y., Henfridsson, O., & Lyytinen, K. (2010). Research Commentary: The New Organizing Logic of Digital Innovation: An Agenda for Information Systems Research. Information Systems Research, 21(4), 724–735. 

Liste der möglichen konkreten Themen:

• SUST-BA-1-1, Sommersemester 2026, Betreuung: Daniel Courtney , M.Sc.

  Creating Novel Data in Digital Healthcare: The Role of Data Complementarities and Intermediaries

  This topic covers how data complementarities enable the creation of new data and new information resources in digital healthcare. In Information Systems research, data complementarities describe situations in which the joint use of multiple resources creates more value than their isolated use. In data-driven environments, combining heterogeneous datasets, infrastructures, and analytics capabilities can not only increase the value of existing data but also lead to the emergence of novel data objects. 

  The integration of these heterogeneous data sources often requires data intermediaries, such as digital platforms, data-sharing infrastructures, health information exchanges, or data marketplaces to facilitate the collection, aggregation and redistribution of data across organizational boundaries. It can therefore be suggested that such intermediaries play a central role in data ecosystems, as they enable complementarities between otherwise disconnected datasets and actors. However, it remains unclear how data intermediaries influence the emergence of new datasets and under which conditions complementarities between data sources can be realized in highly regulated environments such as healthcare. 

  Potential Research Questions: 

  • How does the presence of data intermediaries change data-sharing and data integration practices?
  • How do data intermediaries enable complementarities between different health data sources?
  • How do data ecosystems support the emergence of new data through intermediary actors?

  Literatur

  • Alaimo, C. (2022). From people to objects: The digital transformation of fields. Organization Studies, 43(7), 1091-1114.
  • Alaimo, C., & Kallinikos, J. (2022). Organizations decentered: Data objects, technology and knowledge. Organization Science, 33(1), 19-37.
  • Constantinides, P., Henfridsson, O., & Parker, G. G. (2018). Introduction—platforms and infrastructures in the digital age. Information systems research, 29(2), 381-400.
  • Jacobides, M. G., Cennamo, C., & Gawer, A. (2024). Externalities and complementarities in platforms and ecosystems: From structural solutions to endogenous failures. Research Policy, 53(1), 104906.
  • Jacobides, M. G., Cennamo, C., & Gawer, A. (2018). Towards a theory of ecosystems. Strategic management journal, 39(8), 2255-2276.
  • Ritala, P., & Karhu, K. (2023). Capturing value from data complementarities: A multi-level framework. In Research handbook on digital strategy (pp. 273-288). Edward Elgar Publishing.

• SUST-BA-1-2, Sommersemester 2026, Betreuung: Daniel Courtney , M.Sc.

  Unlocking Value from Health Data: New Forms of Value Creation

  This topic covers how novel value is created from data in digital healthcare environments. In Information Systems research, data is increasingly viewed as a strategic resource that enables innovation when combined with digital technologies, analytics capabilities, and organizational processes. Rather than creating value on its own, data generates value when it is used to enable new services, new decision-making processes, or new business models. In complex digital ecosystems, value often emerges from the interaction of multiple actors, technologies, and data sources.

  In the healthcare context, studying data-driven value creation holds particular interest in today’s day and age, as the integration of secondary data usage is becoming more realized leading to greater amounts of heterogeneous data, including electronic health records, sensor and wearable device data, imaging data, and patient-generated data. The potential use and recombination of these data sources can enable new forms of value creation, but at the same time, realizing such value requires new capabilities for data integration, governance, and analytics. This provides real-time interest in the examination of the processes for how data can enable novel forms of value creation in digital healthcare ecosystems while simultaneously contextualizing it against the observable technological, organizational, and ecosystem level mechanisms required to transform that data into new services, insights, or innovations. 

  Potential Research Questions: 

  • How do emerging data types enable new forms of value?
  • How is value from data created in digital healthcare ecosystems involving multiple actors?
  • How does the recombination of data sources lead to new forms of value in healthcare?
  • How does data-driven innovation reshape value creation in healthcare information systems? 

  Literatur

  • Aranyossy, M., & Halmosi, P. (2024). Healthcare 4.0 value creation–The interconnectedness of hybrid value propositions. Technological Forecasting and Social Change, 208, 123718.
  • Ghosh, K., Dohan, M. S., Veldandi, H., & Garfield, M. (2023). Digital transformation in healthcare: insights on value creation. Journal of Computer Information Systems, 63(2), 449-459.
  • Hlongwane, S., & Grobbelaar, S. S. (2022). A practical framework for value creation in health information systems from an ecosystem perspective: Evaluated in the South African context. Frontiers in Psychology, 13, 637883.
  • Kilgus, T., Patecka, A., Schurig, T., Kari, A., Gubser, R., Gersch, M., ... & Fürstenau, D. (2024). Creating value from the secondary use of health data: International examples, best practices, and opportunities to scale. Communications of the Association for Information Systems, 55(1), 507-534.
  • Konopik, J. (2023). The impact of digital platforms and ecosystems in healthcare on value creation—a integrative review and research agenda. IEEE access, 11, 135811-135819.
  • Lim, C., Kim, K. H., Kim, M. J., Heo, J. Y., Kim, K. J., & Maglio, P. P. (2018). From data to value: A nine-factor framework for data-based value creation in information-intensive services. International journal of information management, 39, 121-135.
  • Nienstedt, J., & Trenz, M. (2025). From sharing to profiting: Exploring the interplay between value creation and strategic appropriation in data ecosystems. Electronic Markets, 35(1), 1-24.

This topic examines digital transformation and AI adoption in established non-digital SMEs, focusing on the mechanisms, success factors, barriers, and impacts on business models. SMEs face unique organizational, technological, and environmental challenges when implementing digital and AI-driven initiatives. In this context, physical or manufacturing SMEs are those primarily engaged in producing tangible goods through manufacturing or assembly, relying on physical resources, labour and equipment to create products with material value. The topic explores how organizational capabilities, resources, and external enablers influence digital transformation and AI adoption, and how these, in turn, shape business practices, value creation, and operational processes. Students will apply theoretical perspectives, including dynamic capabilities and external enabler frameworks, to analyse patterns of technology adoption, innovation, and opportunity emergence in SMEs.

Literatur

• Aghazadeh, Hashem; Zandi, Farzad; Amoozad Mahdiraji, Hannan; Sadraei, Razieh (2024): Digital transformation and SME internationalisation: unravelling the moderated-mediation role of digital capabilities, digital resilience and digital maturity. In: Journal of Enterprise Information Management 37 (5), S. 1499–1526. DOI: 10.1108/JEIM-02-2023-0092.

• Canhoto, Ana Isabel; Quinton, Sarah; Pera, Rebecca; Molinillo, Sebastián; Simkin, Lyndon (2021): Digital strategy aligning in SMEs: A dynamic capabilities perspective. In: The Journal of Strategic Information Systems 30 (3), S. 101682. DOI: 10.1016/j.jsis.2021.101682.

• Li, Liang; Su, Fang; Zhang, Wei; Mao, Ji‐Ye (2018): Digital transformation by SME entrepreneurs: A capability perspective. In: Information Systems Journal 28 (6), S. 1129–1157. DOI: 10.1111/isj.12153.

• Peretz-Andersson, Einav; Tabares, Sabrina; Mikalef, Patrick; Parida, Vinit (2024): Artificial intelligence implementation in manufacturing SMEs: A resource orchestration approach. In: International Journal of Information Management 77, S. 102781. DOI: 10.1016/j.ijinfomgt.2024.102781

Liste der möglichen konkreten Themen:

• SUST-BA-2-1, Sommersemester 2026, Betreuung: Ann Christin Conrady , MIB MSM

  Success Factors and Barriers for Manufacturing SME Digital Transformation

  This topic examines how digital transformation enables manufacturing SMEs to thrive. It focuses on organizational capabilities, leadership, resource availability and external ecosystem support as key enablers, as well as barriers such as limited digital skills, financial constraints and resistance to change. Using a dynamic capability perspective, students will review the literature to explore how SMEs sense digital opportunities, seize them through strategic action and adapt their business models, value creation and operational processes to create value. AI adoption may serve as one possible tool within broader digital transformation initiatives, supporting innovation and process improvement where relevant. The topic highlights recurring patterns of success and failure, identifies critical success factors and barriers and provides practical insights for SMEs seeking to leverage digital technologies for growth and competitiveness.

  Research Questions:

  • How do organizational capabilities, resources, and external ecosystem support enable manufacturing SMEs to leverage digital transformation for growth and competitiveness?
  • Which success factors and barriers have the greatest impact on SMEs thriving through digital transformation and business model adaptation?

  Literatur

  • Canhoto, Ana Isabel; Quinton, Sarah; Pera, Rebecca; Molinillo, Sebastián; Simkin, Lyndon (2021): Digital strategy aligning in SMEs: A dynamic capabilities perspective. In: The Journal of Strategic Information Systems 30 (3), S. 101682. DOI: 10.1016/j.jsis.2021.101682.
  • Costa, Eric; Soares, António Lucas; Sousa, Jorge Pinho de (2020): Industrial business associations improving the internationalisation of SMEs with digital platforms: A design science research approach. In: International Journal of Information Management 53, S. 102070. DOI: 10.1016/j.ijinfomgt.2020.102070.
  • Li, Liang; Su, Fang; Zhang, Wei; Mao, Ji‐Ye (2018): Digital transformation by SME entrepreneurs: A capability perspective. In: Information Systems Journal 28 (6), S. 1129–1157. DOI: 10.1111/isj.12153.
  • Merhi, Mohammad I. (2023): An evaluation of the critical success factors impacting artificial intelligence implementation. In: International Journal of Information Management 69, S. 102545. DOI: 10.1016/j.ijinfomgt.2022.102545.

  • Soluk, Jonas; Kammerlander, Nadine (2021): Digital transformation in family-owned Mittelstand firms: A dynamic capabilities perspective. In: European Journal of Information Systems 30 (6), S. 676–711. DOI: 10.1080/0960085X.2020.1857666.

  • Polisetty, Aruna; Chakraborty, Debarun; G, Sowmya; Kar, Arpan Kumar; Pahari, Subhajit (2024): What Determines AI Adoption in Companies? Mixed-Method Evidence. In: Journal of Computer Information Systems 64 (3), S. 370–387. DOI: 10.1080/08874417.2023.2219668.

  • Rana, Nripendra P.; Chatterjee, Sheshadri; Dwivedi, Yogesh K.; Akter, Shahriar (2022): Understanding dark side of artificial intelligence (AI) integrated business analytics: assessing firm’s operational inefficiency and competitiveness. In: European Journal of Information Systems 31 (3), S. 364–387. DOI: 10.1080/0960085X.2021.1955628.

• SUST-BA-2-2, Sommersemester 2026, Betreuung: Ann Christin Conrady , MIB MSM

  External Enablers of Opportunity Emergence in Physical SMEs

  This topic examines whether digital technologies and sustainability initiatives enable new opportunities to emerge for physical SMEs. It focuses on the role of external enablers, including digital tools, networks and environmental developments, in facilitating opportunity recognition and organizational change. Drawing on the external enabler perspective, students will review how digital and sustainable initiatives can hinder or support opportunity emergence in SMEs, how these opportunities vary in their potential and the mechanisms through which external enablers influence business growth. The topic highlights both the enabling potential and the non-deterministic nature of opportunity generation, with implications for SME strategy and practice.

  Research Questions:

  • How do digital technologies and other external enablers create opportunities for physical SMEs to innovate or expand?
  • Under which conditions do external enablers lead to successful opportunity realization and organizational change in physical SMEs?

  Literatur

  • Davidsson, P., Recker, J., & von Briel, F. (2022). External enablers of entrepreneurship. In Oxford research encyclopedia of business and management.

  • Von Briel, F., Davidsson, P., & Recker, J. (2018). Digital technologies as external enablers of new venture creation in the IT hardware sector. Entrepreneurship Theory and Practice, 42(1), 47-69.

  • von Briel, F., Davidsson, P., & Recker, J. (2025). Why and how societal crises give rise to extreme growth outliers: A theory of external enablement. Academy of Management Review, (ja), amr-2023.

  • von Briel, F., Davidsson, P., & Recker, J. (2025). External Enablement Is Non-Deterministic but the Occurrence of Extreme Growth Outliers Is Still Systematic: A Reply to “Outliers by Accident” and “Outliers in Waiting”. Academy of Management Review, (ja), amr-2025.

  • von Briel, F., Recker, J., & Davidsson, P. (2018). Not all digital venture ideas are created equal: Implications for venture creation processes. The Journal of Strategic Information Systems, 27(4), 278-295.

Recent measurement reports estimate that a substantial proportion of internet activity is automated. The Imperva Bad Bot Report 2025 estimates that 51% of global web traffic in 2024 originated from automated agents. Beyond automation, research has also documented large-scale bot participation in online communication (Ferrara et al., 2016). This concern has also been reflected in the speculative concept known as the “Dead Internet Theory”, which suggests that a growing share of online content—including social interactions, articles, and comments—could be generated by bots or AI systems rather than humans, creating an environment where distinguishing genuine human activity from synthetic content becomes increasingly difficult (Ahmed et al., 2024; Schmitt & Flechais, 2024; Lopez-Joya et al., 2025).

Traditional bots were template-based and rule-driven (Ferrara et al., 2016); repetitive and timing-regular (Subrahmanian et al., 2016); detectable via stylometric, temporal, and graph features (Varol et al., 2017). However, modern AI-powered bots can contextually interpret posts and adapt responses (Brown et al., 2020), generate semantically coherent, tailored comments (Feng et al., 2024), produce multimodal content (text + images) via generative systems (Ramesh et al., 2022), embed themselves into network structures that mimic social communities (Peng et al., 2024), coordinate persuasive narratives across swarms of accounts (Pacheco et al., 2020; Lopez-Joya et al., 2025). 

In this seminar, students will examine the social implications of AI-generated discourse and the technical challenges of detecting increasingly human-like bots, asking:

“At what point do LLM-powered bot networks challenge the detectability and integrity of human discourse in online platforms?”

The seminar connects the following:

• Botnet architectures

• Dead Internet Theory

• Platform detection tradeoffs

• social implications of AI

• Adversarial AI

Literatur

• Ahmed, A., Qamar, R., Asif, R., Imran, M., & Ahmed, S. (2024). The Dead Internet Theory: Investigating the Rise of AI-Generated Content and Bot Dominance in Cyberspace. Pakistan Journal of Engineering, Technology & Science (PJETS), 12(1), 37-48.

• Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., ... & Amodei, D. (2020). Language models are few-shot learners. Advances in neural information processing systems, 33, 1877-1901.

• Feng, S., Wan, H., Wang, N., Tan, Z., Luo, M., & Tsvetkov, Y. (2024, August). What does the bot say? opportunities and risks of large language models in social media bot detection. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

• Ferrara, E., Varol, O., Davis, C., Menczer, F., & Flammini, A. (2016). The rise of social bots. Communications of the ACM, 59(7), 96–104.

• Imperva. (2025). 2025 Bad bot report.

• Karim, A., Salleh, R. B., Shiraz, M., Shah, S. A. A., Awan, I., & Anuar, N. B. (2014). Botnet detection techniques: review, future trends, and issues. Journal of Zhejiang University-SCIENCE C (Computers & Electronics), 15(11), 943-983.

• Li, D., Shen, L., Guo, Q., Zhang, C., Li, J., Jiang, W., & Yu, M. (2025). BotLGT: Social bot detection based on LLM and graph transformer. Neurocomputing, 131453.

• Lopez-Joya, S., Diaz-Garcia, J. A., Ruiz, M. D., & Martin-Bautista, M. J. (2025). Dissecting a social bot powered by generative AI: anatomy, new trends and challenges. Social Network Analysis and Mining, 15(1), 7.

• Pacheco, D., Hui, P. M., Torres-Lugo, C., Truong, B. T., Flammini, A., & Menczer, F. (2020). Uncovering coordinated networks on social media. arXiv preprint arXiv:2001.05658, 16.

• Peng, H., Zhang, J., Huang, X., Hao, Z., Li, A., Yu, Z., & Yu, P. S. (2024). Unsupervised social bot detection via structural information theory. ACM Transactions on Information Systems, 42(6), 1-42.

• Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022). Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125, 1(2), 3.

• Schmitt, M., & Flechais, I. (2024). Digital deception: Generative artificial intelligence in social engineering and phishing. Artificial Intelligence Review, 57(12), 324.

• Subrahmanian, V. S., Azaria, A., Durst, S., Kagan, V., Galstyan, A., Lerman, K., ... & Menczer, F. (2016). The DARPA Twitter bot challenge. Computer, 49(6), 38-46.

Liste der möglichen konkreten Themen:

• SUST-BA-3-1, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  Who Runs the Web? – Bots!

  While early bots were simple scripts that posted repetitive messages (Klopfenstein et al., 2017), modern bots increasingly use large language models (LLMs) and generative AI to produce realistic content, participate in conversations, and coordinate persuasive narratives across networks (Adam et al., 2025; McTear et al., 2024). Bots no longer just spam—they simulate social consensus (Lindner et al., 2024).

  Students explore how generative AI enables large-scale synthetic persuasion systems to understand how bots influence discourse and coordinate networks of accounts to simulate agreement and impact online discussions (Radivojevic, Clark & Brenner, 2024).

  Building on propaganda network analysis (Benkler et al. 2018; Chaudhari & Pawar, 2021) and coordinated behavior networks (Pacheco et al., 2020; Magelinski et al., 2022), students may examine the modern architecture of synthetic persuasion systems:

  General Bot Persuasion Pipeline

  1. Operator defines narrative objective
  2. LLM generates persuasive framing
  3. Multimodal model generates visual reinforcement
  4. Bot swarms distribute coordinated/adaptive posts
  5. Engagement amplification (likes, replies)
  6. Counter-skeptic response and narrative persistence

  Moreover, empirical studies show that perceived consensus significantly increases diffusion speed and belief adoption (Cai et al., 2022; Marchegiani, B., 2025). LLM-powered coordination may amplify this effect by simulating distributed agreement at scale (Lopez-Joya et al., 2025). Students can also investigate how simulated consensus may influence belief adoption and information diffusion.

  Potential Research Questions (RQs)

  RQ1 – Does the integration of large language models into social bot networks make the “Dead Internet” scenario technically plausible?

  RQ2 – How do modern generative bots differ from earlier social bots? How has LLM integration altered the persuasive capacity of social bots compared to pre-transformer systems? 

  RQ3 – What multimodal capabilities most enhance persuasive realism? (optional) Does contextual adaptation increase persuasion success?

  RQ4 – Can simulated consensus increase/amplify belief adoption or information diffusion? 

  Literatur

  • Adam, M., Bauer, K., Jussupow, E., Stein, M. K., & Benlian, A. (2025). Generating Tomorrow’s Me: How Collaborating with Generative AI Changes Humans: M. Adam et al. Business & Information Systems Engineering, 67(5), 583-594.

  • Benkler, Y., Faris, R., & Roberts, H. (2018). Network propaganda: Manipulation, disinformation, and radicalization in American politics. Oxford University Press.

  • Cai, M., Luo, H., Meng, X., & Cui, Y. (2022). Differences in behavioral characteristics and diffusion mechanisms: A comparative analysis based on social bots and human users. Frontiers in Physics, 10, 875574.

  • Chaudhari, D. D., & Pawar, A. V. (2021). Propaganda analysis in social media: a bibliometric review. Information Discovery and Delivery, 49(1), 57-70.

  • Klopfenstein, L. C., Delpriori, S., Malatini, S., & Bogliolo, A. (2017, June). The rise of bots: A survey of conversational interfaces, patterns, and paradigms. In Proceedings of the 2017 conference on designing interactive systems (pp. 555-565).

  • Lindner, I., Heidergott, B., Badri, S., & Praum, M. (2024). The impact of bots on social learning and consensus formation: Why even an “infinitesimal” number of bots matters. Available at SSRN 4851284.

  • Lopez-Joya, S., Diaz-Garcia, J. A., Ruiz, M. D., & Martin-Bautista, M. J. (2025). Dissecting a social bot powered by generative AI: anatomy, new trends and challenges. Social Network Analysis and Mining, 15(1), 7.

  • Pacheco, D., Hui, P. M., Torres-Lugo, C., Truong, B. T., Flammini, A., & Menczer, F. (2021, May). Uncovering coordinated networks on social media: methods and case studies. In Proceedings of the international AAAI conference on web and social media (Vol. 15, pp. 455-466).

  • Magelinski, T., Ng, L., & Carley, K. (2022). A synchronized action framework for detection of coordination on social media. Journal of Online Trust and Safety, 1(2).

  • Marchegiani, B. (2025). Anthropomorphism, false beliefs, and conversational AIs: how chatbots undermine users' autonomy. Journal of Applied Philosophy, 42(5), 1399-1419.

  • McTear, M., & Ashurkina, M. (2024). Transforming conversational AI: Exploring the power of large language models in interactive conversational agents. Springer Nature.

  • Radivojevic, K., Clark, N., & Brenner, P. (2024, May). Llms among us: Generative ai participating in digital discourse. In Proceedings of the AAAI symposium series (Vol. 3, No. 1, pp. 209-218).

• SUST-BA-3-2, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  Blurred Lines & the Adversarial Arms Race Between Bots and Platforms

  Early bot detection systems relied heavily on metadata anomalies, network clustering patterns, and behavioral irregularities (Karim et al., 2014). Traditional bots often exhibited identifiable signatures such as repetitive content, regular posting intervals, or tightly connected clusters within social networks (Xiao et al., 2015). However, advances in generative AI allow modern AI-driven bots to increasingly evade traditional classifiers (Li et al., 2025; Peng et al., 2024).

  On top of that, recent adversarial analyses suggest that the integration of large language models (LLMs) into online bots has introduced an arms-race dynamic between automated agents and platform detection systems, where generative capability and detection robustness continuously co-evolve (Feng et al., 2024).

  This escalation creates a arms-race dynamic between bots and platforms:

  1. Bots become increasingly humanlike through generative models
  2. Platforms tighten detection thresholds
  3. Detection systems begin producing more false positives
  4. Legitimate users are mistakenly flagged or removed
  5. Moderation and enforcement costs increase for platforms

  As a result, bot detection can be understood as a tradeoff problem under adversarial pressure and resource constraints. This dynamic can be analyzed through signal detection theory, which predicts a fundamental relationship between sensitivity (detecting bots) and specificity (avoiding false positives) under uncertainty (Green & Swets, 1966). Increasing detection sensitivity inevitably increases the likelihood that legitimate users will be misclassified. In attention-driven digital markets, excessive false positives may also impose economic costs, as removing legitimate users reduces engagement and advertising revenue (Kärki, K. 2024; Cennamo & Karanovic, 2025).

  Under sustained adversarial escalation, platforms may converge toward a stable equilibrium where bots cannot be fully eliminated, some level of automation is economically tolerated while enforcement strictness remains strategically bounded. From this perspective, the “Dead Internet” hypothesis can be reframed not as a conspiracy theory but as a theoretical question about adversarial scalability, statistical indistinguishability, and platform economics.

  Potential Research Questions (RQs)

  RQ1 – Does adversarial escalation between bots and platforms impose fundamental limits on large-scale bot detection?

  RQ2 – How does the integration of large language models change the detectability of social bots compared to earlier bot systems?

  RQ3 – What behavioral, network, or multimodal signals remain effective for distinguishing LLM-powered bots from human users?

  Literatur

  • Cennamo, C., & Karanovic, J. (2025). Are Social Media Platforms a Threat to Democracy? An Ecosystem Governance Perspective. Journal of Management Studies.

  • Green, D. M., & Swets, J. A. (1966). Signal detection theory and psychophysics (Vol. 1, pp. 1969-2012). New York: Wiley.

  • Karim, A., Salleh, R. B., Shiraz, M., Shah, S. A. A., Awan, I., & Anuar, N. B. (2014). Botnet detection techniques: review, future trends, and issues. Journal of Zhejiang University-SCIENCE C (Computers & Electronics), 15(11), 943-983.

  • Kärki, K. (2024). Digital distraction, attention regulation, and inequality. Philosophy & Technology, 37(1), 8.

  • Li, D., Shen, L., Guo, Q., Zhang, C., Li, J., Jiang, W., & Yu, M. (2025). BotLGT: Social bot detection based on LLM and graph transformer. Neurocomputing, 131453.

  • Peng, H., Zhang, J., Huang, X., Hao, Z., Li, A., Yu, Z., & Yu, P. S. (2024). Unsupervised social bot detection via structural information theory. ACM Transactions on Information Systems, 42(6), 1-42.

  • Xiao, C., Freeman, D. M., & Hwa, T. (2015, October). Detecting clusters of fake accounts in online social networks. In Proceedings of the 8th ACM Workshop on Artificial Intelligence and Security (pp. 91-101).

• SUST-BA-4, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  AI Agentic Cyberspace Attacks: Prompt Injection, Sleeper Agents, and the Security of Skill Ecosystems

  As AI systems evolve from passive chatbots into tool-using agents, they increasingly interact with external services such as APIs, file systems, browsers, plugins, and execution environments. These integrations expand the functionality of AI systems but also introduce new security risks. Researchers increasingly describe these emerging attacks, where vulnerabilities can arise through third-party skills, external data sources, or tool integrations (Deng et al., 2025; Wang et al., 2025).

  Recent studies highlight several emerging threats in these ecosystems. Prompt injection attacks can manipulate agent behavior by inserting adversarial instructions into external content such as webpages or documents (Geng et al., 2026; Gulyamov et al., 2026). In addition, sleeper-agent attacks embed instructions that remain dormant until specific triggers—such as keywords, contextual cues, or time-based events—activate them (Souri et al., 2022). These mechanisms resemble delayed-execution malware or trojanized dependencies in traditional software systems (Duarte et al., 2026; Geng et al., 2026). A further risk arises from credential leakage, where exposed API keys or authentication tokens allow attackers to access external infrastructure through compromised agents (Shi et al., 2025).

  This seminar examines AI agents as autonomous execution environments rather than passive text generators — and asks: “Are we recreating the worst vulnerabilities of the software supply chain inside LLM ecosystems?”

  Literatur

  • Deng, Z., Guo, Y., Han, C., Ma, W., Xiong, J., Wen, S., & Xiang, Y. (2025). Ai agents under threat: A survey of key security challenges and future pathways. ACM Computing Surveys, 57(7), 1-36.

  • Duarte, J. D., Cândido, G. D., De Britto Filho, J. R. A., Neto, J. S., Costa, E. J., Da Costa, J. P. J., & De Melo, L. P. (2026). A Systematic Review of Prompt Injection Attacks on Large Language Models: Trends, Taxonomy, Evaluation, Defenses and Opportunities. IEEE Access.

  • Gulyamov, S., Rodionov, A., Khursanov, R., Mekhmonov, K., Babaev, D., & Rakhimjonov, A. (2026). Prompt Injection Attacks in Large Language Models and AI Agent Systems: A Comprehensive Review of Vulnerabilities, Attack Vectors, and Defense Mechanisms. Information, 17(1), 54.

  • Geng, T., Xu, Z., Qu, Y., & Wong, W. E. (2026). Prompt Injection Attacks on Large Language Models: A Survey of Attack Methods, Root Causes, and Defense Strategies. Computers, Materials & Continua, 87(1).

  • Shi, Y., Yang, Z., Zhong, K., Yang, G., Yang, Y., Zhang, X., & Yang, M. (2025). The Skeleton Keys: A Large Scale Analysis of Credential Leakage in Mini-apps. In NDSS.

  • Souri, H., Fowl, L., Chellappa, R., Goldblum, M., & Goldstein, T. (2022). Sleeper agent: Scalable hidden trigger backdoors for neural networks trained from scratch. Advances in Neural Information Processing Systems, 35, 19165-19178.

  • Wang, Y., Pan, Y., Guo, S., & Su, Z. (2025). Security of internet of agents: Attacks and countermeasures. IEEE Open Journal of the Computer Society.

As AI systems evolve from passive chatbots into tool-using agents, they increasingly interact with external services such as APIs, file systems, browsers, plugins, and execution environments. These integrations expand the functionality of AI systems but also introduce new security risks. Researchers increasingly describe these emerging attacks, where vulnerabilities can arise through third-party skills, external data sources, or tool integrations (Deng et al., 2025; Wang et al., 2025).

Recent studies highlight several emerging threats in these ecosystems. Prompt injection attacks can manipulate agent behavior by inserting adversarial instructions into external content such as webpages or documents (Geng et al., 2026; Gulyamov et al., 2026). In addition, sleeper-agent attacks embed instructions that remain dormant until specific triggers—such as keywords, contextual cues, or time-based events—activate them (Souri et al., 2022). These mechanisms resemble delayed-execution malware or trojanized dependencies in traditional software systems (Duarte et al., 2026; Geng et al., 2026). A further risk arises from credential leakage, where exposed API keys or authentication tokens allow attackers to access external infrastructure through compromised agents (Shi et al., 2025).

This seminar examines AI agents as autonomous execution environments rather than passive text generators — and asks: “Are we recreating the worst vulnerabilities of the software supply chain inside LLM ecosystems?”

Literatur

• Deng, Z., Guo, Y., Han, C., Ma, W., Xiong, J., Wen, S., & Xiang, Y. (2025). Ai agents under threat: A survey of key security challenges and future pathways. ACM Computing Surveys, 57(7), 1-36.

• Duarte, J. D., Cândido, G. D., De Britto Filho, J. R. A., Neto, J. S., Costa, E. J., Da Costa, J. P. J., & De Melo, L. P. (2026). A Systematic Review of Prompt Injection Attacks on Large Language Models: Trends, Taxonomy, Evaluation, Defenses and Opportunities. IEEE Access.

• Gulyamov, S., Rodionov, A., Khursanov, R., Mekhmonov, K., Babaev, D., & Rakhimjonov, A. (2026). Prompt Injection Attacks in Large Language Models and AI Agent Systems: A Comprehensive Review of Vulnerabilities, Attack Vectors, and Defense Mechanisms. Information, 17(1), 54.

• Geng, T., Xu, Z., Qu, Y., & Wong, W. E. (2026). Prompt Injection Attacks on Large Language Models: A Survey of Attack Methods, Root Causes, and Defense Strategies. Computers, Materials & Continua, 87(1).

• Shi, Y., Yang, Z., Zhong, K., Yang, G., Yang, Y., Zhang, X., & Yang, M. (2025). The Skeleton Keys: A Large Scale Analysis of Credential Leakage in Mini-apps. In NDSS.

• Souri, H., Fowl, L., Chellappa, R., Goldblum, M., & Goldstein, T. (2022). Sleeper agent: Scalable hidden trigger backdoors for neural networks trained from scratch. Advances in Neural Information Processing Systems, 35, 19165-19178.

• Wang, Y., Pan, Y., Guo, S., & Su, Z. (2025). Security of internet of agents: Attacks and countermeasures. IEEE Open Journal of the Computer Society.

Liste der möglichen konkreten Themen:

• SUST-BA-4-1, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  Prompt Injection and Sleeper Agents in Tools using LLMs

  Prompt injection attacks occur when adversarial instructions manipulate a model into executing unintended actions. In agent-based systems, these instructions may be embedded in external resources such as webpages, documents, or third-party skills, which the agent processes automatically (Duarte et al., 2026; Geng et al., 2026).

  A more advanced variant involves “sleeper” instructions that remain dormant until specific triggers occur, such as keywords, contextual cues, or time-based events (Hubinger et al., 2024). These attacks resemble established cybersecurity techniques such as logic bombs or trojanized dependencies (Deng et al., 2025).

  As agents increasingly interact with external content and tools, prompt injection may function similarly to software supply-chain attacks, where malicious components compromise downstream systems (Zhang et al., 2025).

  Potential Research Questions (RQs)

  RQ1 – Are current agent frameworks fundamentally insecure by design?

  RQ2 – How do prompt injection attacks differ from traditional code injection?

  RQ3 – Can delayed-trigger (“sleeper”) instructions be reliably detected before execution?

  RQ4 – What architectural safeguards (e.g., sandboxing or capability restrictions) reduce agent exploitability?

  Literatur

  • Deng, Z., et al. (2025). AI agents under threat: A survey of key security challenges and future pathways. ACM Computing Surveys.

  • Duarte, J. D., et al. (2026). A systematic review of prompt injection attacks on large language models. IEEE Access.

  • Geng, T., et al. (2026). Prompt injection attacks on large language models: Root causes and defense strategies. Computers, Materials & Continua.

  • Hubinger, E., Denison, C., Mu, J., Lambert, M., Tong, M., MacDiarmid, M., ... & Perez, E. (2024). Sleeper agents: Training deceptive llms that persist through safety training. arXiv preprint arXiv:2401.05566.

  • Zhang, J., Bu, H., Wen, H., Liu, Y., Fei, H., Xi, R., ... & Meng, D. (2025). When llms meet cybersecurity: A systematic literature review. Cybersecurity, 8(1), 55.

• SUST-BA-4-2, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  Token Leakage, Credential Abuse and Compromise

  AI agents often rely on API keys, authentication tokens, and external service credentials to interact with tools and cloud services. These secrets may be exposed through prompts, logs, plugin configurations, or external repositories (Wang et al., 2025).

  Credential leakage enables attackers to access external infrastructure, perform unauthorized actions, or extract sensitive data. These risks resemble traditional supply-chain compromises, where leaked credentials or malicious dependencies enable system-wide attacks (Deng et al., 2025).

  Because conversational interfaces encourage informal interaction and agents may automatically access external tools, managing credentials securely becomes a critical challenge for AI agent ecosystems.

  Potential Research Questions (RQs)

  RQ1 – What threat models best describe supply-chain attacks in AI agent ecosystems?

  RQ2 – Should AI agents operate under zero-trust assumptions when accessing external tools?

  RQ3 – How should credentials be securely handled in AI agent architectures? and what architectural patterns prevent privilege escalation after credential leaks?

  RQ4 – Can LLM systems detect accidental credential disclosure during interactions?

  Literatur

  • Deng, Z., et al. (2025). AI agents under threat: A survey of key security challenges and future pathways. ACM Computing Surveys.

  • Rabzelj, M., & Sedlar, U. (2025). Beyond the leak: Analyzing the real-world exploitation of stolen credentials using honeypots. Sensors, 25(12), 3676.

  • Shi, Y., Yang, Z., Zhong, K., Yang, G., Yang, Y., Zhang, X., & Yang, M. (2025). The Skeleton Keys: A Large Scale Analysis of Credential Leakage in Mini-apps. In NDSS.

  • Wang, Y., Pan, Y., Guo, S., & Su, Z. (2025). Security of internet of agents: Attacks and countermeasures. IEEE Open Journal of the Computer Society.

Large language models (LLMs) have recently demonstrated strong performance on tasks that require reasoning, including mathematical problem solving, logical inference, and multi-step planning (Wei et al., 2022; Kojima et al., 2022; Yao et al., 2023). These capabilities often emerge when models generate intermediate reasoning steps through prompting strategies such as such as chain-of-thought, program-of-thought, or tree-of-thought reasoning frameworks, which encourages models to decompose complex problems into sequential reasoning steps. Empirical studies show that such prompting significantly improves performance on arithmetic, commonsense reasoning benchmarks demonstrating intelligence (Chen, W. et al., 2023; Gao et al., 2022; Yao et al., 2023; Wei et al., 2022).

However, there is ongoing debate about whether these systems genuinely reason or instead perform sophisticated statistical pattern completion learned during training (Bender et al., 2021; Chen, Q. et al., 2025; Mitchell, 2021). Critics argue that strong benchmark performance may reflect pattern recognition rather than structured reasoning processes (Eriksson et al., 2025), while proponents suggest that reasoning abilities may emerge from model scaling and improved prompting strategies (Wei et al., 2022; Yao et al., 2023).

A central difficulty in this debate is the absence of a clear, falsifiable definition of reasoning. Without such a definition, it becomes difficult to determine whether any AI system truly possess reasoning capabilities or whether apparent reasoning behavior reflects statistical correlations learned during training. Some researchers therefore argue that reasoning should be operationalized and empirically evaluated through carefully designed tasks and benchmarks.

This seminar examines how reasoning in machine systems can be defined, measured, and empirically evaluated. Students will explore how reasoning relates to abstraction, compositional generalization, and systematicity and whether current benchmarks meaningfully evaluate these capabilities.

This seminar theme addresses the core philosophical and empirical problem: 

“What is reasoning? And what objective test would conclusively demonstrate it in current AI systems?”

Literatur

• Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? FAccT.

• Chen, W., Ma, X., Wang, X., & Cohen, W. W. (2023). Program of thoughts prompting: Disentangling computation from reasoning in large language models.

• Chen, Q., Qin, L., Liu, J., Peng, D., Guan, J., Wang, P., ... & Che, W. (2025). Towards reasoning era: A survey of long chain-of-thought for reasoning large language models. arXiv preprint arXiv:2503.09567.

• Eriksson, M., Purificato, E., Noroozian, A., Vinagre, J., Chaslot, G., Gomez, E., & Fernandez-Llorca, D. (2025, October). Can we trust ai benchmarks? an interdisciplinary review of current issues in ai evaluation. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (Vol. 8, No. 1, pp. 850-864).

• Gao, L., Madaan, A., Zhou, S., et al. (2022). PAL: Program-aided language models.

• Mitchell, M. (2021). Why AI is harder than we think. arXiv.

• Wei, J., Wang, X., Schuurmans, D., et al. (2022). Chain-of-thought prompting elicits reasoning in large language models. NeurIPS.

• Yao, S., Yu, D., Zhao, J., et al. (2023). Tree of thoughts: Deliberate problem solving with large language models.

Liste der möglichen konkreten Themen:

• SUST-BA-5-1, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  The Leaderboard Illusion

  A key challenge in evaluating LLM reasoning is defining it in a way that is objective, measurable, and falsifiable (Mitchell, 2021). Many current benchmarks measure task performance but may not isolate reasoning abilities from memorization or pattern matching  (Eriksson et al., 2025; Singh et al., 2025).

  Researchers have proposed evaluating reasoning through tasks that require abstraction, compositional generalization, and transfer across domains. The Abstraction and Reasoning Corpus (ARC) was designed to measure these abilities by presenting novel problems that cannot easily be solved through memorized patterns (Chollet, 2019). Other benchmarks by DeepL such as BIG-Bench Hard and GSM8K test whether models can perform multi-step reasoning or mathematical inference (Cobbe, 2021; Suzgun, 2023).

  Recent work suggests that prompting techniques such as chain-of-thought prompting can significantly improve reasoning performance by encouraging models to generate intermediate reasoning steps (Wei et al., 2022). However, it remains debated whether these outputs reflect genuine reasoning or structured pattern retrieval learned during training (Bender et al., 2021; Chen, Q. et al., 2025).

  Students will investigate how reasoning can be empirically tested and whether current evaluation methods and benchmarks successfully distinguish reasoning from statistical generalization.

  Potential Research Questions (RQs)

  RQ1 – Are current benchmarks reliable proxies for evaluating reasoning or intelligence?

  RQ2 – How can reasoning be distinguished from statistical pattern completion in LLMs?

  RQ3 – Does chain-of-thought prompting like techniques reveal genuine reasoning processes or structured pattern retrieval?

  Literatur

  • Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? FAccT.

  • Chollet, F. On the measure of intelligence. arXiv preprint arXiv:1911.01547, doi.org/10.48550/arXiv.1911.01547 (2019).

  • Chen, Q., Qin, L., Liu, J., Peng, D., Guan, J., Wang, P., ... & Che, W. (2025). Towards reasoning era: A survey of long chain-of-thought for reasoning large language models. arXiv preprint arXiv:2503.09567.

  • Cobbe, K., Kosaraju, V., Bavarian, M., Chen, M., Jun, H., Kaiser, L., ... & Schulman, J. (2021). Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.

  • Eriksson, M., Purificato, E., Noroozian, A., Vinagre, J., Chaslot, G., Gomez, E., & Fernandez-Llorca, D. (2025, October). Can we trust ai benchmarks? an interdisciplinary review of current issues in ai evaluation. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (Vol. 8, No. 1, pp. 850-864).

  • Mitchell, M. (2021). Why AI is harder than we think. arXiv.

  • Singh, S., Nan, Y., Wang, A., D'souza, D., Kapoor, S., Üstün, A., ... & Hooker, S. (2025). The leaderboard illusion. arXiv preprint arXiv:2504.20879.

  • Suzgun, M., Scales, N., Schärli, N., Gehrmann, S., Tay, Y., Chung, H. W., ... & Wei, J. (2023, July). Challenging big-bench tasks and whether chain-of-thought can solve them. In Findings of the Association for Computational Linguistics: ACL 2023 (pp. 13003-13051).

  • Wei, J., et al. (2022). Chain-of-thought prompting elicits reasoning in large language models. NeurIPS.

• SUST-BA-5-2, Sommersemester 2026, Betreuung: Mahnoor Shahid , M.Sc.

  Every Time AI Gets Smart, We Move the Goalposts

  Throughout the history of artificial intelligence, systems have repeatedly achieved high performance on tasks that were once considered indicators of intelligence, such as playing chess, mastering Go, or recognizing objects in images (Silver et al., 2016; Krizhevsky et al., 2012). However, once machines succeed at these tasks, they are often reinterpreted as “mere computation” rather than intelligence (Marcus & Davis, 2020; Mitchell, 2021).

  This phenomenon raises an important epistemological question: are definitions of intelligence inherently unstable? If intelligence is defined by tasks that machines cannot perform, then each technological breakthrough may shift the criteria used to evaluate intelligence.

  Students will examine how benchmark success influences perceptions of intelligence and whether current AI evaluation methods suffer from a moving-target problem. 

  Potential Research Questions (RQs)

  RQ1 – What formal definition of intelligence can be empirically tested in machine systems? 

  RQ2 – Why do definitions of intelligence often change after AI systems achieve benchmark success?

  Literatur

  • Chollet, F. (2019). On the measure of intelligence. arXiv.

  • Hoffmann, C. H. (2022). Is AI intelligent? An assessment of artificial intelligence, 70 years after Turing. Technology in Society, 68, 101893.

  • Mitchell, M. (2023). AI’s challenge problem: Understanding vs. imitation. Communications of the ACM.

  • Pili, G. (2019). Toward a philosophical definition of intelligence. The International Journal of Intelligence, Security, and Public Affairs, 21(2), 162-190.

  • Warner, M. (2019). Wanted: A definition of ‘intelligence’. In Secret Intelligence (pp. 4-12). Routledge.

  • Wheaton, K. J., & Beerbower, M. T. (2006). Towards a new definition of intelligence. Stan. L. & Pol'y Rev., 17, 319.

Digital transformation is one of the dominant phenomena of our time and affects organizations across industries (Hanelt et al., 2021; Kraus et al., 2022). Profound changes, fueled by the increasing use of digital technologies across all areas of life, are taking place at all levels and are transforming organizations, industries, and society as a whole (Ismail et al., 2017; Kraus et al., 2022). Despite the high level of attention in academia and practice, digital transformation, as a complex and multifaceted phenomenon, remains incompletely understood, with heterogeneous conceptualizations and definitions (Vial, 2019; Wessel et al., 2025). 

Multiple drivers of digital transformation processes, such as disruptive changes triggered by the increasing adoption of digital technologies, are discussed in the literature, as well as a broad set of positive as well as negative outcomes of digital transformation, such as increased efficiency, but also privacy issues (Vial, 2019). Due to the interdisciplinary nature and heterogeneous conceptualizations, the state of research is still fragmented (Hanelt et al., 2021; Kraus et al., 2022).

Gaining a deeper understanding of contemporary micro-level but also large-scale digital transformation phenomena is crucial for ensuring long-term success in digital transformation processes and engages academia and practitioners alike.

Literatur

• Hanelt, A., Bohnsack, R., Marz, D., & Antunes Marante, C. (2021). A systematic review of the literature on digital transformation: Insights and implications for strategy and organizational change. Journal of management studies, 58(5), 1159-1197.
• Kraus, S., Durst, S., Ferreira, J. J., Veiga, P., Kailer, N., & Weinmann, A. (2022). Digital transformation in business and management research: An overview of the current status quo. International journal of information management, 63, 102466.
• Ismail, M. H., Khater, M., & Zaki, M. (2017). Digital business transformation and strategy: What do we know so far. Cambridge Service Alliance, 10(1), 1-35.
• Vial, G. (2019). Understanding digital transformation: A review and a research agenda. The Journal of Strategic Information Systems, 28(2), 118-144.
• Wessel, L., Mosconi, E., Indulska, M., & Baiyere, A. (2025). Digital Transformation: Quo Vadit? Information Systems Journal, 35(4), 1294-1308.

Liste der möglichen konkreten Themen:

• TM-BA-1-1, Sommersemester 2026, Betreuung: Isabella Urban , M. Sc.

  Managing Digital Transformation: A Paradox Theory Perspective

  Paradox theory describes organizational tensions as persistent, simultaneously occurring contradictions that must be addressed productively rather than resolved (Schad et al., 2016). At its core is an integrative "both-and" paradigm that allows for the simultaneous pursuit of seemingly opposing demands such as stability and change (Smith & Lewis, 2011). This perspective is particularly crucial in managing digital transformation, as organizations must combat issues such as ensuring efficiency within existing structures while simultaneously driving disruptive innovation (Kraus et al., 2022; Verhoef et al., 2021). Paradox theory thus provides a fruitful conceptual framework for gaining a deeper understanding of these ambivalences and for successfully navigating and sustaining digital transformation in the long term.

  Against this background, semi-structured interviews with practitioners will be conducted to identify key paradoxical tensions organizations face and to gain further insights into how organizations cope with these.

  Literatur

  • Kraus, S., Durst, S., Ferreira, J. J., Veiga, P., Kailer, N., & Weinmann, A. (2022). Digital transformation in business and management research: An overview of the current status quo. International journal of information management, 63, 102466.
  • Schad, J., Lewis, M. W., Raisch, S., & Smith, W. K. (2016). Paradox research in management science: Looking back to move forward. Academy of management annals, 10(1), 5-64.
  • Smith, W. K., & Lewis, M. W. (2011). Toward a theory of paradox: A dynamic equilibrium model of organizing. Academy of management Review, 36(2), 381-403.
  • Verhoef, P. C., Broekhuizen, T., Bart, Y., Bhattacharya, A., Dong, J. Q., Fabian, N., & Haenlein, M. (2021). Digital transformation: A multidisciplinary reflection and research agenda. Journal of business research, 122, 889-901.

• TM-BA-1-2, Sommersemester 2026, Betreuung: Isabella Urban , M. Sc.

  Responsible Digital Transformation: A Paradox Theory Perspective

  Paradox theory conceptualizes organizational tensions as persistent, interwoven contradictions that cannot simply be resolved but must be integrated (Schad et al., 2016). It advocates a "both-and" approach that enables the simultaneous management of competing demands, thereby fostering both flexibility and stability (Smith & Lewis, 2011). This is particularly relevant in the context of responsible digital transformation and digital responsibility, as organizations must balance technological innovation, economic efficiency, and regulatory compliance with issues such as ethical responsibility and social legitimacy (Mihale-Wilson et al., 2022; Recker et al., 2025). Paradox theory potentially acts as a theoretical lens, to gain a deeper understanding on the resulting paradoxical tensions and how organizations can strategically manage these while simultaneously ensuring sustainable, responsible digital value creation.

  Against this background, a systematic analysis of the literature will be conducted to identify key paradoxical tensions that arise in navigating digital responsibility and responsible digital transformation. Based on this, implications for practice and for further research will be outlined.

  Literatur

  • Mihale-Wilson, C., Hinz, O., van der Aalst, W., & Weinhardt, C. (2022). Corporate digital responsibility: Relevance and opportunities for business and information systems engineering. Business & Information Systems Engineering, 64(2), 127-132.
  • Recker, J., Chatterjee, S., Sundermeier, J., & Tarafdar, M. (2025). Digital responsibility: Current perspectives and future directions. Journal of the Association for Information Systems, 26(5), 1222-1238.
  • Schad, J., Lewis, M. W., Raisch, S., & Smith, W. K. (2016). Paradox research in management science: Looking back to move forward. Academy of management annals, 10(1), 5-64.
  • Smith, W. K., & Lewis, M. W. (2011). Toward a theory of paradox: A dynamic equilibrium model of organizing. Academy of management Review, 36(2), 381-403.

• TM-BA-1-3, Sommersemester 2026, Betreuung: Isabella Urban , M. Sc.

  The Interplay of Organizational and Industry-level Digital Transformation

  Digital transformation is taking place not only at the level of individual organizations but also at the level of entire industries, creating a dynamic interplay between organizational strategies and industry-specific developments (Plekhanov et al., 2021; Vial, 2019). Organizations must align their internal transformation initiatives with market structures, technological standards, and competitive dynamics to remain relevant (Mann et al., 2022; Warner & Wäger, 2029). At the same time, collective industry innovations and platforms shape the opportunities and limitations of individual transformation projects. Understanding this interplay is crucial for strategically managing transformation, leveraging synergies, and addressing risks arising from external changes early on.

  Against this background, a systematic analysis of the literature will be conducted to identify key elements of the interplay between organizational- and industry-level digital transformation, and implications for managing digital transformation will be derived.

  Literatur

  • Mann, G., Karanasios, S., & Breidbach, C. F. (2022). Orchestrating the digital transformation of a business ecosystem. The Journal of Strategic Information Systems, 31(3), 101733.
  • Plekhanov, D., Franke, H., & Netland, T. H. (2023). Digital transformation: A review and research agenda. European management journal, 41(6), 821-844.
  • Vial, G. (2019). Understanding digital transformation: A review and a research agenda. The Journal of Strategic Information Systems, 28(2), 118-144.
  • Warner, K. S., & Wäger, M. (2019). Building dynamic capabilities for digital transformation: An ongoing process of strategic renewal. Long range planning, 52(3), 326-349.

• TM-BA-1-4, Sommersemester 2026, Betreuung: Ali Ergün

  The Influence of Charisma in Digital Transformation?

  While leadership is widely recognized as essential for digital DT success, various studies only superficially indicate how and which leadership behaviors are necessary to achieve expected transformation results. In current literature reviews, Vial (2019) as well as Müller et al. (2024) especially call for more research on the role of organizational actors in DT highlighting that managerial action is key to transformation success in times of high DT dynamics. 

  Having understood charismatic leadership theory through rigorous development and validation in the domains of applied psychology and organizational behavior over the last 30 years, a great deal of research indicates its importance for IS-driven transformation efforts (e.g., Neufeld et al., 2007; Romm & Pliskin, 1999; Wang et al., 2022). However, an explicit analysis and understanding of leadership styles in detail, in particular of charisma and charismatic leadership in the context of DT, is yet missing: A sound understanding of the impact and value of charismatic leadership behavior in digital transformation and its different processes is missing. Considering the significance of charisma on followers in IS discipline as well as discipline-overarching presented by Ergün & Plattfaut (2024), we aim to shed light on this phenomenon by experimentally investigating charismatic leadership behavior in the context of DT processes.

  As part of this seminar paper, 3 to 5 (semi-structured) interviews with organizational members (experienced with DT) have to be conducted to identify the influence of charisma / charismatic leadership on DT success/performance.

  Literatur

  • Ergün, A. N. & Plattfaut, R. (2025). Charismatic leadership behavior DT processes: an interdisciplinary review. ECIS 2025 Proceedings, 9.
  • Markus, M. L., & Rowe, F. (2021). Guest Editorial: Theories of Digital Transformation: A Progress Report. Journal of the Association for Information Systems, 22(2), 273–280.
  • Müller, S. D., Konzag, H., Nielsen, J. A., & Sandholt, H. B. (2024). Digital transformation leadership competencies: A contingency approach. International Journal of Information Management, 75, 102734.
  • Neufeld, D. J., Dong, L., & Higgins, C. (2007). Charismatic leadership and user acceptance of information technology. European Journal of Information Systems, 16(4), 494–510.
  • Romm, C., & Pliskin, N. (1999). The role of charismatic leadership in diffusion and implementation of e mail. Journal of Management Development, 18(3), 273–291.
  • Rowe, F., & Markus, M. L. (2023). Envisioning Digital Transformation: Advancing Theoretical Diversity. Journal of the Association for Information Systems, 24(6), 1459–1478.
  • Vial, G. (2019). Understanding digital transformation: A review and a research agenda. Journal of Strategic Information Systems, 28, 118–144.
  • Wang, W. T., Luo, M. C., & Chang, Y. M. (2022). Exploring the Relationship between Conflict Management and Transformational Leadership Behaviors for the Success of ERP Customization. Information Systems Management, 39(2), 177–200
  • Weber, E., Büttgen, M., & Bartsch, S. (2022). How to take employees on the digital transformation journey: An experimental study on complementary leadership behaviors in managing organizational change. Journal of Business Research, 143, 225–238.
  • Wessel, L., Baiyere, A., Ologeanu-Taddei, R., Cha, J., & Blegind Jensen, T. (2021). Unpacking the Difference Between Digital Transformation and IT-Enabled Organizational Transformation. Journal of the Association for Information Systems, 22(1), 102–129. 

• TM-BA-1-5, Sommersemester 2026, Betreuung: Ali Ergün

  Managerial and Technological responsibilities in Digital Transformation Phases

  DT describes “a process where digital technologies create disruptions triggering strategic responses from organizations that seek to alter their value creation paths while managing the structural changes and organizational barriers that affect the positive and negative outcomes of this process” (Vial, 2019, p. 118). Although theoretical plurality exists on the start, its unfolding, and the effects of DT (Rowe & Markus, 2023), most theoretical perspectives on DT share the view that coordinated action across various management levels is crucial. 

  Contemporary research highlights this actor-oriented coordination and interaction challenge, calling for more research on socio-technical microfoundations for DT, i.e., the role of individual organizational actors in DT (Iden & Bygstad, 2025; Vial, 2019; Wiener et al., 2025) with a “particular focus on their time-related perceptions and actions, in large businesses (where such positions are more likely to exist and be clearly defined)” (Wiener et al., 2025, p. 18).

  In this context, how actors with different roles interact across time and context (i.e., DT phases), remains inadequately theorized (Wiener et al., 2025), leaving open how organizational actors lead, manage, contribute and interact with each other in order to jointly digitally transform their organization. By examining DT as a phase-based and actor-driven phenomenon, this seminar paper investigates DT as a sequence of evolving interactions among organizational actors who must continuously coordinate across domains of expertise. In particular, this qualitative study explores how technological and managerial responsibilities are distributed across DT phases, and how these distributions shape outcomes.

  As part of this seminar paper, 4 (semi-structured) interviews with organizational members (experienced with DT) from technological (e.g., CDO) and managerial roles have to be conducted to identify how roles and responsibilities interact, coordinate, and align for successful DT in different DT phases.

  Literatur

  • Iden, J., & Bygstad, B. (2025). Sociotechnical micro-foundations for digital transformation. European Journal of Information Systems, 34(2), 367–382.
  • Karabag, S. F., Simonsson, J., Berggren, C., Andreasson, M., & Eriksson, R. (2026). Digital transformation as a multi-phase process: A longitudinal study of corporate strategy and business unit adaptation. Journal of Business Research, 202, 115796.
  • Rowe, F., & Markus, M. L. (2023). Envisioning Digital Transformation: Advancing Theoretical Diversity. Journal of the Association for Information Systems, 24(6), 1459–1478.
  • Vial, G. (2019). Understanding digital transformation: A review and a research agenda. Journal of Strategic Information Systems, 28, 118–144.
  • Wiener, M., Strahringer, S., & Kotlarsky, J. (2025). Where are the processes in IS research on digital transformation? A critical literature review and future research directions. The Journal of Strategic Information Systems, 34(2), 101900. 

• TM-BA-1-6, Sommersemester 2026, Betreuung: Jannis Nacke

  Structured Literature Review on Data-Driven Organizations

  Becoming a data-driven organization refers to the systematic use of data to inform decisions, shape strategies, and guide operational activities. Instead of relying primarily on intuition, organizations increasingly aim to embed data analysis into everyday processes and routines (Fischer et al. 2023; Kotlarsky et al. 2024).In this context, data is treated as a strategic resource that enables improved decision quality, process optimization, and enhanced organizational performance (Brown 2020; Aral et al. 2012). Despite these potential benefits, many organizations struggle to translate data initiatives into sustained impact. Prior research suggests that becoming data-driven involves more than the implementation of analytical tools and technologies and instead requires broader organizational changes.

  The aim of this seminar paper is to conduct a structured literature review on data-driven organizations. Students are expected to systematically analyze existing research to identify key challenges associated with becoming data-driven and how organizations address these challenges. The review should focus on uncovering recurring patterns in the literature, including typical barriers, enabling factors, and organizational responses (e.g., capabilities, practices, or mechanisms) that support the successful adoption of data-driven approaches. 

  Students are encouraged to synthesize and structure the literature, develop their own categorization of challenges and responses, and highlight similarities and differences across studies. The goal is to provide a structured overview of the current state of research and to derive implications for both research and organizational practice.

  Literatur

  • Aral, Sinan; Brynjolfsson, Erik; Wu, Lynn (2012): Three-Way Complementarities: Performance Pay, Human Resource Analytics, and Information Technology. In Management Science 58 (5), pp. 913–931.
  • Brown, Sara (2020): How to build a data-driven company. In MIT Sloan School of Management.
  • Fischer, Hannes; Wiener, Martin; Strahringer, Susanne; Kotlarsky, Julia; Bley, Katja (2023): Data-Driven Organizations: Review, Conceptual Framework, and Empirical Illustration. In Australasian Journal of Information Systems 27.
  • Kotlarsky, Julia; Oshri, Ilan; Sarker, Suprateek (2024): The Bumpy Road to Becoming a Data-Driven Enterprise. In CAIS 55 (1), pp. 193–204. 

Having understood the significant effects of artificial intelligence (AI) on operational efficiency (Cui et al., 2024), organizations strategically focus on deploying and providing artificial intelligence in their processes and structures. Organizational members across departments now routinely and increasingly interact with conversational AI to reduce highly repetitive tasks and focus more on higher-level cognitive tasks at work, with 88% of AI users coming from non-technical professions (De Smet et al, 2023). AI interactions are likely to become even more common as generative AI is increasingly used in roles and functions previously reserved for humans, such as HR, IT, finance, or customer service and support (Tey et al, 2024). Estimations of the World Economic Forum indicate that by 2020, AI-driven automation will autonomously take over one-third of all work tasks (Di Battista et al., 2025). 

The use of GenAI brings implications for the working environment that are important today and in the future in order to generate hoped-for efficiencies: the way of working through the use of AI in business processes is changing and interpersonal collaboration in teams is being influenced as AI is used alongside human colleagues for monitoring, coordination and operational work. With AI being increasingly embedded in collaborative processes, this technology challenges, the understanding of the technology itself (Larson & DeChurch, 2020), traditional notions of teamwork (Richter & Schwabe, 2025), and intragroup processes (Zercher et al. 2023). 

Literatur

• Cui, K. Z., Demirer, M., Jaffe, S., Musolff, L., Peng, S., & Salz, T. (2026). The effects of generative AI on high-skilled work: Evidence from three field experiments with software developers. Management Science.
• De Smet, A., Durth, S., Hancock, B., Baldocchi, M., & Reich, A. (2023). The human side of generative AI: Creating a path to productivity. McKinsey & Company
• Di Battista, A., Grayling, S., Játive, X., Leopold, T., Li, R., Sharma, S., & Zahidi, S. (2025). Future of jobs report 2025. In World Economic Forum, Geneva, Switzerland.
• Larson, L., & DeChurch, L. A. (2020). Leading teams in the digital age: Four perspectives on technology and what they mean for leading teams. The leadership quarterly, 31(1), 101377.
• Richter, A., & Schwabe, G. (2025). “There is No ‘AI’in ‘TEAM’! Or is there?”–Towards meaningful human-AI collaboration. Australasian Journal of Information Systems, 29.
• Tey, K. S., Mazar, A., Tomaino, G., Duckworth, A. L., & Ungar, L. H. (2024). People judge others more harshly after talking to bots. PNAS nexus, 3(9), pgae397.
• Zercher, D., Jussupow, E., & Heinzl, A. (2023). When AI joins the Team: A Literature Review on Intragroup Processes and their Effect on Team Performance in Team-AI Collaboration. ECIS 2023 Research Papers. 307. 

Liste der möglichen konkreten Themen:

• TM-BA-2-1, Sommersemester 2026, Betreuung: Isabella Urban , M. Sc.

  Managing Artificial Intelligence: A Paradox Theory Perspective

  Managing artificial intelligence “marks the dawn of a new age of information technology management” (Berente et al., 2021) but capturing and sustaining value from artificial intelligence appropriation is a key challenge for organizations across industries (Collins et al., 2021). With the rise of artificial intelligence, numerous paradoxical tensions arise that organizations need to combat, e.g., the tension between augmentation and automation (Raisch & Krakowski, 2021). Paradox Theory analyzes how organizations can manage simultaneously existing, conflicting demands without abandoning one goal at the expense of another (Smith & Lewis, 2011). It promotes a "both-and" approach that enables the pursuit of stability and change, as well as control and innovation, concurrently. In the context of artificial intelligence, this is particularly relevant, as companies must balance issues such as artificial intelligence efficiency and performance with flexibility, adaptability, and the pressure to innovate. Paradox Theory provides a theoretical framework for gaining a deeper understanding of these multidimensional tensions and for promoting successful AI appropriation in the long term.

  Against this background, semi-structured interviews with practitioners will be conducted to identify key paradoxical tensions organizations face and to gain further insights into how organizations cope with these.

  Literatur

  • Berente, N., Gu, B., Recker, J., & Santhanam, R. (2021). Managing artificial intelligence. MIS quarterly, 45(3), 1433-1450.
  • Collins, C., Dennehy, D., Conboy, K., & Mikalef, P. (2021). Artificial intelligence in information systems research: A systematic literature review and research agenda. International journal of information management, 60, 102383.
  • Raisch, S., & Krakowski, S. (2021). Artificial intelligence and management: The automation–augmentation paradox. Academy of management review, 46(1), 192-210.
  • Schad, J., Lewis, M. W., Raisch, S., & Smith, W. K. (2016). Paradox research in management science: Looking back to move forward. Academy of management annals, 10(1), 5-64.
  • Smith, W. K., & Lewis, M. W. (2011). Toward a theory of paradox: A dynamic equilibrium model of organizing. Academy of management Review, 36(2), 381-403.

• TM-BA-2-2, Sommersemester 2026, Betreuung: Ali Ergün

  The Impact of AI on (Individual) Learning

  Artificial intelligence is increasingly reshaping work, learning, and professional development. This is particularly relevant for early-career workers and students entering the labour market. Recent findings show declining employment opportunities for young workers in areas such as customer service and software development since 2022 (Brynjolfsson et al., 2025). At the same time, jobs and tasks exposed to generative AI are especially affected by these developments (Eloundou et al., 2023).

  A key reason is the rapid progress of AI capabilities. Modern AI systems already outperform humans in a growing range of domains and continue to improve steadily (Maslej et al., 2025). As a result, AI is not only augmenting human work but also increasingly automating tasks that have traditionally served as learning opportunities for juniors. This applies to simple tasks, but increasingly also to more complex ones. In addition, generative AI is reshaping work itself by changing how individuals structure and adapt their tasks through job crafting (Mayer et al., 2025). These developments raise important questions: How will juniors learn to become seniors if entry-level tasks are increasingly automated?

  As part of this seminar paper, a systematic literature review will be conducted to assess the state of research in the context of AI and its impact on learning.

  Literatur

  • Bruhin, O., Ebel, P., & Bauer-H, I. (2025). Unlocking Business Value with Generative AI: Balancing Co-Creation and Co-Destruction in Knowledge Work. ECIS 2025.
  • Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at work. The Quarterly Journal of Economics, 140(2), 889-942.
  • Diebel, C., Goutier, M., Adam, M., & Benlian, A. (2025). The Price of AI Assistance: The Undermining Effect of AI-Generated Code on Developers’ Procedural Knowledge. ECIS 2025.
  • Eloundou, T., Manning, S., Mishkin, P., & Rock, D. (2024). GPTs are GPTs: Labor market impact potential of LLMs. Science, 384(6702), 1306-1308.
  • Maslej, N., Fattorini, L., Perrault, R., Gil, Y., Parli, V., Kariuki, N., ... & Oak, S. (2025). Artificial intelligence index report 2025. arXiv preprint arXiv:2504.07139.
  • Mayer, T., & Schwehn, T.-J. (2025). Moving Beyond Task Efficiency: How Generative AI Challenges Teamwork. ECIS 2025.

• TM-BA-2-3, Sommersemester 2026, Betreuung: Dr. Carolin Vollenberg

  Adoption of Artificial Intelligence in Organizations

  The rapid technological development in the case of Generative AI (GenAI) has fundamentally altered the technological landscape, presenting organizations with unprecedented opportunities for value creation. However, making GenAI effectively usable for companies requires a shift in focus from purely technical capabilities to practical, organizational implementation. This process is not merely defined by what is technologically feasible, but is influenced by complex social and organizational factors.

  Given the diversity of modern organizations, GenAI gives rise to a vast spectrum of use cases that demand specific integration and scaling strategies. Consequently, identifying the specific challenges—ranging from cultural resistance to structural inertia—and the successful strategies for adoption is critical. This seminar paper will therefore conduct a systematic analysis of existing literature to evaluate which challenges organizations face in the context of GenAI adoption and how to potentially overcome these challenges. 

  Literatur

  • Jung, J. Y., & Winter, S. J. (2025). Policy for Sociotechnical Gen AI Assessment: Leveraging End Users. Journal of the Association for Information Systems, 26(2), 287–293.
  • Kemell, K.‑K., Saarikallio, M., Nguyen-Duc, A., & Abrahamsson, P. (2025). Still just personal assistants? – A multiple case study of generative AI adoption in software organizations. Information and Software Technology, 186, 107805.
  • Seifdar, M. H., & Amiri, B. (2025). Strategic adoption of generative AI in organizations: A game-theoretic and network-based approach. International Journal of Information Management, 84, 102932.
  • Shao, J., Ahmad, H., Kamal, M. M., Butt, A. H., Zhang, J. Z., & Alam, F. (2025). Unveiling the potential: exploring the adoption of GenAI and its impact on organizational outcomes. Journal of Managerial Psychology. Advance online publication.
  • Yavetz, G., & Nakash, M. (2026). Adopting GenAI applications in the workplace: managerial implications and insights from ICT professionals. EuroMed Journal of Business, 1–19. 

• TM-BA-2-4, Sommersemester 2026, Betreuung: Dr. Carolin Vollenberg

  Adoption of Artificial Intelligence in Organizations (empirical)

  The rapid technological development in the case of Generative AI (GenAI) has fundamentally altered the technological landscape, presenting organizations with unprecedented opportunities for value creation. However, making Artificial Intelligence (AI) effectively usable for companies requires a shift in focus from purely technical capabilities to practical, organizational implementation. This process is not merely defined by what is technologically feasible, but is heavily influenced by complex social and organizational factors.

  Given the diversity of modern organizations, GenAI gives rise to a vast spectrum of use cases that demand specific integration and scaling strategies. Consequently, identifying the specific challenges—ranging from cultural resistance to structural inertia—and the successful strategies for adoption is critical. For this paper, the goal is to empirically investigate the challenges that organizations face in the adoption of GenAI and how they build up strategies to overcome them. 

  Literatur

  • Nieken, P. (2023). Charisma in the gig economy: The impact of digital leadership and communication channels on performance. The Leadership Quarterly, 34(6), 101631.
  • Jung, J. Y., & Winter, S. J. (2025). Policy for Sociotechnical Gen AI Assessment: Leveraging End Users. Journal of the Association for Information Systems, 26(2), 287–293.
  • Kemell, K.‑K., Saarikallio, M., Nguyen-Duc, A., & Abrahamsson, P. (2025). Still just personal assistants? – A multiple case study of generative AI adoption in software organizations. Information and Software Technology, 186, 107805.
  • Seifdar, M. H., & Amiri, B. (2025). Strategic adoption of generative AI in organizations: A game-theoretic and network-based approach. International Journal of Information Management, 84, 102932.
  • Shao, J., Ahmad, H., Kamal, M. M., Butt, A. H., Zhang, J. Z., & Alam, F. (2025). Unveiling the potential: exploring the adoption of GenAI and its impact on organizational outcomes. Journal of Managerial Psychology. Advance online publication.
  • Yavetz, G., & Nakash, M. (2026). Adopting GenAI applications in the workplace: managerial implications and insights from ICT professionals. EuroMed Journal of Business, 1–19. 

As digitalization increasingly permeates both professional environments and everyday life, the rapid speed of technological advancement has created a critical societal and organizational challenge: the risk of leaving significant portions of the population behind. Therefore, it is important to explore what factors for digital exclusion are and how it is possible to overcome the hereby given challenges. Digital inclusion is defined not only by access to digital technology but also by the competencies to use it effectively (Schabram et al., 2023). Thus, it is not only the hardware access that is relevant for digital inclusion, but also opportunities to use digital technology and thereby build competencies, for example, at their workplace or through resources offered by other institutions, such as libraries. To bridge the digital divide, it is therefore necessary to address specific fears and barriers and to promote the development of digital skills. One possibility to foster collective learning in the context of new digital technologies is the cultivation of Communities of Practice. Communities of practice are “groups of people who share a concern, a set of problems, or a passion about a topic, and who deepen their knowledge and expertise in this area by interacting on an ongoing basis" (Wenger et al., 2002). These communities can provide an anchor for individuals navigating digital transitions both at the workplace and in broader social contexts.

Literatur

• Davenport, E., & Hall, H. (2002). Organizational Knowledge and Communities of Practice. Annual Review of Information Science and Technology (ARIST), 36, 171–227.
• Helsper, E. J. (2008). Digital inclusion: an analysis of social disadvantage and the information society. Department for Communities and Local Government.
• Schabram, G., Schulze, K., & Stilling, G. (2023). Armut und digitale Teilhabe:Empirische Befunde zur Frage des Zugangs zur digitalen Teilhabe in Abhängigkeit von Einkommensarmut. Kurzexpertise der Paritätischen Forschungsstelle.
• Wenger, E., McDermott, R., & Snyder, W. M. (2002). Cultivating communities of practice: A guide to managing knowledge. Harvard Business School Press.

Liste der möglichen konkreten Themen:

• TM-BA-3-1, Sommersemester 2026, Betreuung: Ronja Rieger

  Communities of Practice

  Organizations face the constant challenge of managing rapid knowledge turnover and fostering effective collective learning to remain competitive. To address this, many organizations leverage Communities of Practice (CoPs)—informal or formal groups that foster collective learning —as a vehicle for knowledge exchange and innovation. This literature analysis aims to explore how organizations strategically apply or foster communities of practice to drive innovation and facilitate the integration of new technologies. By synthesizing existing research, the paper identifies key success factors and different conceptualizations of CoPs that are frequently utilized in organizations, based on organizational and information systems literature.

  Literatur

  • Davenport, E., & Hall, H. (2002). Organizational Knowledge and Communities of Practice. Annual Review of Information Science and Technology (ARIST), 36, 171–227.
  • Haas, A., Borzillo, S., & Bootz, J.‑P. (2026). Communities of practice and innovation: a review and research agenda. Journal of Knowledge Management, 30(3), 983–1010. 

• TM-BA-3-2, Sommersemester 2026, Betreuung: Ronja Rieger

  Digital Inclusion and Exclusion

  As digitalization increasingly permeates both professional environments and everyday life, the risk of leaving significant portions of the population behind has become a critical societal challenge. This trend carries the potential to further marginalize vulnerable groups, transforming technological advancement into a barrier rather than an opportunity. This literature analysis aims to define the multi-faceted concept of digital inclusion and identify the specific risk factors and systemic barriers that lead to digital exclusion. By synthesizing current research, the paper investigates which socio-economic and technical hurdles most significantly impede participation in the digital sphere. Finally, the study explores potential solutions and access strategies described in the literature as effective means to foster equitable digital engagement and close the growing "digital divide."

  Literatur

  • Helsper, E. J. (2008). Digital inclusion: an analysis of social disadvantage and the information society. Department for Communities and Local Government.
  • Ragnedda, M., Ruiu, M. L., & Addeo, F. (2022). The self-reinforcing effect of digital and social exclusion: The inequality loop. Telematics and Informatics, 72, 101852.
  • van Dijk, J. (2005). The Deepening Divide: Inequality in the Information Society. SAGE Publications, Inc.