Shifting cognitive structures in human-AI collaborative learning design

Abstract

This study examines collaborative learning design with a generative artificial intelligence (GenAI) co-pilot among in-service university English teachers in China. In a between-subjects experiment ( N = 60), epistemic network analysis (ENA) was used to model the situated discourse structures of team dialogues, complemented by thematic analysis of interviews. AI-supported teams achieved higher overall scores on a rubric covering constructive alignment, Universal Design for Learning principles, and other aspects of learning design. ENA identified task-specific differences in collaborative discourse, with AI-supported teams displaying a more interconnected and learner-centered network. Participants described the AI as a pedagogical conscience and an intellectual sparring partner whose prompts kept professional judgment visible during the design process. The findings suggest that, in complex professional tasks, GenAI may support reflection and pedagogical reasoning alongside design work.