Supporting shared mental models: How an AI facilitator shapes interprofessional team cognition in collaborative learning.
Abstract
Background
Effective interprofessional collaboration is critical for patient safety. Fostering the underlying cognitive coordination of teamwork remains a significant challenge in health professions education. Traditional approaches often prioritize outcome assessment rather than actively facilitating the collaborative process itself.
Objectives
This study investigated whether a novel artificial intelligence (AI) facilitator, acting as a metacognitive guide, was associated with team performance and differences in the collaborative discourse structure of interprofessional student teams.
Methods
In an explanatory sequential mixed-methods randomized controlled trial, thirty three-person teams (N = 90) of final-year clinical medicine, nursing, and pharmacy students were randomized to solve a complex clinical case with or without the AI facilitator. While blinded experts assessed the primary outcome of team performance, this study utilized epistemic network analysis (ENA) on collaborative discourse to map discourse-based proxies for cognitive coordination, followed by reflexive thematic analysis of post-task interviews to identify underlying interactional mechanisms.
Results
Teams in the experimental group showed higher clinical performance and reported greater collaboration satisfaction and lower cognitive load. ENA revealed distinct structural differences, with the experimental network showing stronger connections among professional contribution, hypothesis generation, and synthesis and integration. Thematic analysis suggested three perceived interactional mechanisms: functioning as a non-directive cognitive guide, supporting cross-professional perspective-taking, and providing a metacognitive framework that helped organize complex reasoning.
Conclusion
An AI facilitator may support interprofessional teamwork by scaffolding metacognitive regulation during collaboration. These findings suggest that AI may serve as a process-oriented pedagogical scaffold, supporting human interaction and collaborative cognition beyond content delivery.