Enhancing knowledge gain in virtual learning teams: the role of AI explainability, team cohesion, and perceived virtuality

Abstract The integration of Artificial Intelligence (AI) in education has introduced new opportunities for supporting team-based learning in virtual environments. While conversational AI tools like ChatGPT offer scalable and interactive learning support, their effectiveness in enhancing knowledge gain in virtual team settings remains underexplored. This study investigates the role of perceived AI explainability in fostering knowledge gain among learners in AI-supported virtual teams. Drawing on Self-Regulated Learning theory, we propose that AI explainability promotes metacognitive engagement by providing clear, interpretable feedback that scaffolds learners’ planning, monitoring, and reflection processes. We further examine two team-level moderators—Team Perceived Virtuality and Team Cohesion—as contingencies that influence the strength of this relationship. A multi-level, longitudinal study was conducted involving 344 MBA students working in 48 virtual project teams over a 16-week period. Students collaborated with ChatGPT as a team member, and data were collected across three time points. Results reveal that perceived AI explainability significantly enhances knowledge gain, and this relationship is amplified in teams with higher levels of perceived virtuality and cohesion. These findings suggest that the effectiveness of AI in virtual learning contexts is shaped by the team’s social dynamics and the clarity of AI communication. This study contributes to research on AI-enhanced education by emphasizing the importance of explainability and team characteristics in knowledge acquisition and supports the pedagogical design of AI-mediated collaborative learning environments.