Key points are not available for this paper at this time.
Practical instructional skills are crucial for pre-service teachers (PSTs) training. Traditional training methodologies, however, struggle to replicate authentic classroom interactions and are further constrained by ethical considerations regarding the involvement of actual students. To address these challenges, this study proposes an innovative approach using ChatGPT-4o to simulate students to facilitate interactive classroom engagement and provide targeted post-lesson feedback within a controlled training environment. In this setting, zero–shot chain–of-thought prompts guided the AI to generate diverse, realistic student reactions – including probing questions and occasional misunderstandings – to challenge PSTs' adaptive teaching skills. The experimental study involved 29 PSTs from a normal university in eastern China, employing a mixed-methods research design. Quantitative and qualitative analyzes demonstrated that interactions with AI-simulated students significantly enhanced participants' instructional dialogue quality, approximating the proficiency levels observed in experienced educators. Furthermore, participants exhibited marked improvements in self-efficacy and demonstrated increased receptiveness toward AI-mediated training modalities. These empirical findings substantiate the efficacy of AI-based simulation as a scalable and ethically viable alternative in teacher preparation programs, effectively addressing the persistent theory-practice divide in teacher education.
Li et al. (Mon,) studied this question.