Model-based support for teaching practice and self-efficacy in artificial intelligence-enhanced virtual reality teaching simulation

Abstract Effective teaching plays a vital role in fostering cognitive and psychological development in humans. Virtual reality simulations, along with the embedded, artificial intelligence (AI)-powered virtual student agents, offer opportunities for preservice teachers to practice teaching iteratively in a dynamic, context-rich manner. However, preservice teachers face challenges when practicing in these AI-enhanced VR simulations. In this study, we designed and investigated model-based supports aimed at scaffolding preservice teachers during simulation-based pretraining, in-simulation, and post-simulation stages. This experimental study with 57 preservice teachers indicated significant positive impacts of model-based support on teaching knowledge and skills development. Furthermore, significant improvements in knowledge of teaching were observed in the experimental groups using model-based support in the AI-enhanced VR simulation, from pre- to post-test. However, the difference between the experimental and control groups in teaching self-efficacy was nonsignificant. The implications of these findings and potential future directions for designing learning support in VR simulation-based teacher education are discussed.