Reflective AI in Teams

Motivation

The increasing integration of (Gen)AI systems into the workplace is transforming how individuals and groups perform tasks and make decisions. While the understanding of human-AI collaboration at the individual level is more advanced, significantly less attention has been paid to how AI can support teams of individuals. However, a vast share of work in companies is conducted in collaborative, team-like settings, motivating a need to observe AI use in teams.

AI systems may act as moderators in team meetings, assist in coordinating collaboration, support conflict resolution, or even take on leadership-like functions such as task delegation based on team members’ skills. Therefore, AI may introduce benefits such as increased objectivity, potentially mitigating human biases such as favoritism or social dynamics.

However, the use of AI in team settings raises critical challenges. In particular, it is essential to ensure that teams remain in control of decisions and are not passively guided or dominated by AI systems. This calls for mechanisms that enable teams to critically reflect on, understand, and appropriately integrate AI-generated suggestions.

Objectives

This thesis explores how AI systems can support team collaboration while ensuring meaningful human oversight and reflection.

The central goal is to investigate and design reflection mechanisms that enable teams to remain in control when interacting with AI systems. Building on concepts such as Human-in-the-Loop, Explainable AI (XAI), and uncertainty awareness, the thesis examines how these approaches – traditionally applied to individual users – can be adapted and extended to team settings.

Possible research directions include:

  • Identifying and conceptualizing use cases for AI-supported teamwork
  • Designing and prototyping reflection mechanisms that allow teams to critically assess and integrate AI suggestions
  • Empirically investigating how AI suggestions are integrated in team collaboration

The exact focus can be defined together with the supervisors, depending on the student’s interests and level of studies (Bachelor’s/Master’s).

Profile

  • Interest in interdisciplinary research on human-AI collaboration and teamwork
  • Basic understanding of AI/LLMs
  • Programming skills (depending on the chosen focus, especially for prototyping)
  • Self-driven, curious, and open to exploring new research directions
  • Good English skills

Contact

We offer an exciting and highly relevant research topic at the intersection of AI, teamwork, and decision-making, with close supervision and the opportunity to contribute to an emerging research area.

If you are interested, please send your current transcript of records, a short CV, and a brief motivation (2–3 sentences) to Julia Seitz Sänger (julia seitz ∂does-not-exist.kit edu) or Maximilian Förster (maximilian.foerster∂kit.edu).

Starting Literature

Zana Buçinca, Maja Barbara Malaya, and Krzysztof Z. Gajos. 2021. To Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making. Proc. ACM Hum.-Comput. Interact. 5, CSCW1, Article 188 (April 2021), 21 pages. https://doi.org/10.1145/3449287

Hefer, Laura; Gal, Uri; and Hsu, Carol, "AI Meeting Assistants, Views of Organisational Meetings and Potential Implications to Productivity" (2023). ACIS 2023 Proceedings. 112. https://aisel.aisnet.org/acis2023/112

Schulz, T., & Speck, C. (2026). Conceptualising reflective use: Toward a process perspective on human-AI interaction. European Conference on Information Systems. provided on demand by supervisors

Förster, M., Schröppel, P., Schwenke, C., Fink, L., & Klier, M. (2024). Choose Wisely: Leveraging Explainable AI to Support Reflective Decision-Making. International Conference on Information Systems. https://aisel.aisnet.org/icis2024/aiinbus/aiinbus/22/