Problem

Generative AI tools are increasingly becoming part of how students learn, write, research, and solve problems. They can make individual tasks faster and increase output quality by providing explanations, suggestions, summaries, or feedback on demand. At the same time, working with AI introduces a new rhythm into studying. Students formulate prompts, wait for results, inspect generated content, revise their requests, and sometimes start another activity while an AI system is still producing an answer. One central question is what, when and how to delegate to AI and decide what to do while the AI is working (Baird & Maruping, 2021).

These small interruptions may change how students engage with the learning process. Waiting time could provide an opportunity to reflect on a problem, reconsider an approach, or process what has already been learned. However, it could also encourage students to switch to another task, open another AI tool, or divide their attention across several ongoing activities. In this way, AI may increase productivity while simultaneously eroding the independent knowledge and judgment that humans contribute to human-AI collaboration (Fügener et al., 2021).

Scope

In this thesis, you will investigate how the interaction patterns created by AI-assisted studying influence attention, reflection, and learning. The goal is to understand when interruptions and waiting periods during AI use become distracting (i.e., effectively disrupting flow) and when they can instead create useful opportunities for deeper thinking. A useful lens for this is Schön’s (1983) distinction between reflection-in-action and reflection-on-action. Ideally, AI-assisted studying combines both: students reflect-in-action while interacting with the AI, for example by critically examining generated content and adapting their requests, and reflect-on-action during the pauses in which the AI is working, by stepping back to consider what they have done and learned so far. The thesis may also explore how AI interfaces could deliberately support productive pauses rather than continuous task switching.

The thesis will involve reviewing related work on human-AI interaction, delegation to AI, attention and task switching, metacognition, reflection, cognitive offloading, and learning with generative AI; The subsequent empirical work can build on AMSL, the institute’s own generative AI application, which offers a realistic setting for implementing and studying different interaction variants. Methodologically, the thesis can take one of two routes. First, it can test concrete interaction designs in controlled experiments, such as in the KD2Lab, by comparing conditions (e.g., different AI response times, allowing or restricting parallel activities while the AI is working, or introducing explicit reflection periods before students receive or continue working with AI-generated content). Second, following a Design Science Research approach, it can develop and evaluate novel design approaches for AI-assisted studying that deliberately foster reflection-in-action during the interaction and reflection-on-action during pauses.

Potential research questions include how students use periods in which they are waiting for an AI response; whether these periods lead to reflection-on-action or to switching between tasks; whether and how students engage in reflection-in-action while interacting with the AI; how students decide which parts of their work to delegate to the AI; how frequent switching during AI-assisted studying affects perceived flow, cognitive load, and engagement; whether students’ perceived understanding differs from their actual learning; and whether deliberately designed reflection pauses can improve immediate understanding or longer-term knowledge retention.

Suitability

The thesis is suitable for students interested in Human-Computer Interaction, Human-AI Interaction, Human-AI Collaboration, Generative AI, learning technologies, and cognitive psychology.

Experience with empirical user studies, experimental design, Design Science Research, or implementing web-based AI interfaces is helpful, but can also be developed during the thesis.

Main References

Tan, F. F.-Y., Messerschmidt, M. A., Yin, W., & Nov, O. (2026). The Impact of Response Latency and Task Type on Human-LLM Interaction and Perception. Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems.

Tankelevitch, L., Kewenig, V., Simkute, A., Scott, A. E., Sarkar, A., Sellen, A., & Rintel, S. (2024). The Metacognitive Demands and Opportunities of Generative AI. Proceedings of the CHI Conference on Human Factors in Computing Systems.

Barcaui, A. (2025). ChatGPT as a Cognitive Crutch: Evidence from a Randomized Controlled Trial on Knowledge Retention. Social Sciences & Humanities Open, 12, 102287.

Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. Basic Books.

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.

Fügener, A., Grahl, J., Gupta, A., & Ketter, W. (2021). Will Humans-in-the-Loop Become Borgs? Merits and Pitfalls of Working with AI. MIS Quarterly, 45(3), 1527–1556.