Zhihao Zhou, Letao Xiao, Ji Wang, Shuai Tong, Wenxiao Zhang, Ning Hao
ABSTRACT Background The rise of generative artificial intelligence (AI) in educational research has prompted growing interest in how such tools might support collaborative knowledge construction and creative thinking. While most studies focus on individual‐AI interaction, less is known about group‐level cognitive and interactional processes when AI is integrated into shared problem‐solving settings. Objectives This study investigates how group collaboration with AI supports creativity‐relevant processes in collaborative learning contexts and explores the psychological mechanisms that mediate this effect. Methods Using a within‐subject experimental design, 96 university students engaged in a scientific divergent application task under three collaboration modes: group‐AI collaboration, individual‐AI collaboration and human‐only group collaboration. The study adopted a mixed‐method approach, combining behavioural coding of perspective‐taking and AI utilisation strategies with statistical modelling to examine mediation effects. Tasks were designed to simulate open‐ended learning scenarios that emphasise ideation, problem‐solving and conceptual integration—core components of collaborative learning. Results and Conclusions Group‐AI collaboration led to the most creative outputs in terms of usefulness, originality and quality. This effect was mediated by two key mechanisms: strategic use of AI as a generative and exploratory resource and selective, high‐quality integration of both peer and AI‐generated ideas. While these findings stem from a controlled setting, they offer initial evidence that AI‐supported group work may enrich collaborative ideation processes and hold promise for future instructional design and group‐based learning practices.