A Microteaching Training System Integrating Augmented Reality and Generative Artificial Intelligence

Microteaching is widely used to develop teaching skills. However, conventional practices still rely on manual observation and delayed feedback, limiting objectivity and scalability. This study proposes MARApp, a microteaching training system integrating Augmented Reality and Generative Artificial Intelligence to support immersive teaching practice and automated performance evaluation. A one-group pretest - posttest design was conducted with 150 pre-service teachers. Learning outcomes were operationalized as a latent construct reflected by perceived learning effectiveness, knowledge ability, and media interactivity. Data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results indicate improvements in teaching performance, with a strong effect size of learning outcomes on intention to use. Furthermore, learning outcomes significantly influence intention to use both directly and indirectly through performance score, although the mediation effect is relatively small. These findings suggest that integrating immersive environments with automated performance evaluation may support microteaching effectiveness and technology adoption. However, the absence of a control group limits causal interpretation, and future studies should employ more rigorous experimental designs.