How postgraduate learning engagement in generative AI-assisted contexts supports critical thinking: A grounded theory study

Despite growing interest in generative artificial intelligence (GenAI) in higher education, limited attention has been paid to how learning engagement is reconfigured in GenAI-assisted contexts and how it supports critical thinking. This study adopts a grounded theory approach to examine the dimensions of postgraduate learning engagement and its underlying mechanisms. Based on semi-structured interviews with 32 postgraduates, four dimensions are identified: cognitive, emotional, human-AI interaction, and agentic engagement. Extending existing frameworks, the findings show that engagement promotes critical thinking through both direct and indirect pathways. Learning self-efficacy and lower-order cognitive offloading act as key mediators, while GenAI trust serves as a contextual moderator. A process-oriented model is proposed to explain how GenAI reshapes engagement and fosters higher-order thinking.