Towards Pedagogically Structured Human–AI Engagement in a Peer Learning Context

Generative AI is increasingly integrated into higher education, yet prior research has mainly examined AI use at the individual level, with less attention on AI engagement variation across learning tasks, AI roles, and collaborative contexts. This study examined perceived AI usability across a pedagogically structured peer-learning process where generative AI supported different, individual or collaborative learning activities. The context was a compulsory undergraduate course where student teams worked on authentic projects. Responses from 411 students indicate that generative AI was perceived as most usable during collaborative content structuring and least usable during individual reflection. Students’ perceived study ability showed small but consistent associations with perceived AI usability. Similar patterns emerged within teams, suggesting that ways of engaging with generative AI may partly develop as shared practices with study-related self-regulation potentially playing a role. Findings highlight the importance of examining AI engagement as a phenomenon shaped by pedagogical task design, collaborative learning, and the roles assigned to generative AI in learning activities.