Lectures and seminars Seminar: Professor James Skon
Preserving Reasoning in AI-Rich Learning Environments: A Cross-Domain Platform Design and Authoring Workshop
This workshop introduces coLearn-AI, a process-constrained collaborative learning platform originally developed in computer science education to preserve reasoning, dialogue, and epistemic accountability in AI-rich environments. The system integrates structured small-group interaction, rotating roles, enforced turn-taking, and instructor-aligned AI mediation designed to probe thinking without supplying solutions.
The session concludes with discussion of cross-domain adaptation challenges, evaluation strategies, and open research questions concerning responsible AI integration in professional education.
Speaker
Professor James Skon
Professor of Computer Science, Kenyon College
Title
Preserving Reasoning in AI-Rich Learning Environments: A Cross-Domain Platform Design and Authoring Workshop
Hosts
Catharina Hultgren, Sophie Curbo and Annica Lindkvist, Division of Clinical Microbiology, Department of Laboratory Medicine
More about the workshop
Generative AI systems can now produce plausible diagnostic explanations, analytical summaries, and structured reasoning across many domains. While this creates new possibilities for medical and professional education, it also introduces a structural tension: when answers become inexpensive to obtain, instructional designs that focus primarily on outcomes may inadvertently weaken engagement with the reasoning processes that professional expertise requires.
Rather than functioning as an autonomous tutor, the AI operates within explicit instructor-defined boundaries. It may request clarification, surface missing assumptions, or highlight gaps in reasoning, but it does not provide diagnostic answers or bypass core cognitive work. The platform maintains a persistent “epistemic trace” of collaborative discussion, capturing how hypotheses evolve, assumptions are challenged, and conclusions are refined.
Although initially piloted in programming contexts, the underlying architecture is domain-independent. The workshop explores how such a process-constrained framework might be adapted to case-based and diagnostic learning settings, where collaborative interpretation of evidence, differential reasoning, and reflective discussion are central.
In addition to conceptual framing and live demonstration, participants will engage in guided activity development. Working from a sample case or instructional objective, we will collaboratively design a structured, AI-mediated learning activity using the platform’s authoring workflow. This hands-on component focuses on translating domain expertise into AI-constrained scaffolding that supports inquiry rather than shortcutting it.