Abstract
<title>Abstract</title> <p>Addressing the current gap where applications of Generative Artificial Intelligence (GenAI) in education are predominantly theoretical and lack systematic empirical analysis of authentic teaching scenarios, this study investigates the intrinsic models and practical pathways through which GenAI empowers classroom instruction. Employing a grounded theory approach, a systematic analysis was conducted on 65 typical "AI + Higher Education" classroom cases sourced from the official Ministry of Education of China and other online channels. Through a three-level process of open, axial, and selective coding, this research constructed a three-dimensional "Learning Space (Objects)–Teaching Activities (Events)–Role Relationships (People)" model. This model systematically elucidates the intrinsic mechanism by which GenAI empowers the classroom: by providing intelligent tools, optimizing learning objectives, and generating multimodal resources. The study further proposes a cyclical "Pre-class Preparation–In-class Interaction–Post-class Service" practical pathway, offering an operational framework for technology integration. The findings confirm that GenAI is transforming the classroom ecology from a traditional "teacher-student" binary relationship to a "teacher-machine-student" ternary interactive model. In this new model, the teacher's role transitions to that of a learning facilitator, while students become more proactive knowledge explorers, thereby reshaping instructional roles.</p>