Utilizing customized ChatGPT-based chatbots to support preservice teachers’ understanding of collective mathematical argumentation
This study investigated how ChatGPT-based chatbots can support preservice mathematics teachers’ (PMTs) understanding of core teaching practices, specifically collective mathematical argumentation. Drawing on a situated learning perspective and principles of practice-based teacher education, we focused on the first two phases of the Generative Role-play AI Simulation for Pedagogy (GRASP) model, in which we designed customized AI chatbots to function as both knowledge builders and situation generators. Ten PMTs enrolled in a mathematics education course at a U.S. research university engaged with these chatbots to learn foundational ideas of collective argumentation, analyze argument structure, and examine the productivity of the provided classroom scenarios. Data sources included written assignments, reflections, and pre- and post-surveys. Analyses across multiple data sources indicated that most PMTs in this study developed a robust understanding of collective mathematical argumentation and of the teacher moves that facilitate it through their interactions with the customized AI chatbots, while also demonstrating awareness of the chatbots’ usefulness and inherent limitations. These findings highlight both the promise and the constraints of integrating AI tools into teacher preparation and point to design considerations for creating effective AI-supported learning environments.