Exploring Changes in Teaching and Learning through the Development and Implementation of an AI-Based Feedback Model in Physical Education

This study aimed to develop an artificial intelligence (AI) feedback model based on pose estimation technology and to explore changes in teaching and learning processes, as well as their educational meanings, from the perspectives of students and teachers when the model was applied to middle school physical education classes. Using a rapid prototyping approach, a web-based AI feedback model for analyzing basketball shooting and volleyball spiking movements was developed and implemented in actual middle school classes. A qualitative case study was then conducted with three physical education teachers and seven students. Data were collected through in-depth interviews, participant observation, and lesson video recordings, and were analyzed using thematic analysis. The findings are as follows. First, the AI feedback model, utilizing MediaPipe technology, provided real-time joint coordinate extraction, movement classification, and automatic snapshot storage, and was developed as a web-based educational tool that operates without additional software installation. Second, from the learner perspective, students moved away from teacher-centered, one-way feedback and demonstrated a shift toward becoming self-directed learners who monitor and adjust their own movements based on the visual information provided by the AI system. Third, in terms of instructional structure, a triadic interaction system among teacher, student, and AI was observed, and the teacher’s role was reconfigured from a simple feedback provider to a learning designer and facilitator who analyzes learning data and offers individualized guidance. In conclusion, the application of the AI feedback model in physical education classes partially alleviated temporal and spatial constraints associated with individual feedback and demonstrated the potential for providing immediate and individualized feedback. These findings suggest that the AI feedback model can serve as an effective mediating tool for promoting a learner-centered teaching and learning paradigm. This study is meaningful in that it moves beyond discussions focused on technical performance and identifies the practical possibilities of AI in authentic educational contexts.