AI student agents revolutionize how future math teachers train for classrooms

A new AI-powered training tool is changing how future maths teachers prepare for the classroom. Dr. Dabae Lee, an associate professor at Kennesaw State University, has developed virtual student agents to simulate real classroom interactions. Early feedback suggests these simulations feel more realistic than traditional training methods. The system features three AI agents—Gabriel, Noah, and Jiwoo—each designed to represent different learner types with distinct behaviours and thinking styles. Dr. Lee's work builds on Jean Piaget's socio-cognitive conflict theory, which argues that learning grows through social negotiation and perspective-taking. The goal is to help teachers better understand and respond to diverse student reasoning in mathematics.

The tool has been refined through partnerships with institutions like the University of Missouri, using feedback from trials. It focuses on *responsive teaching*—a skill that involves diagnosing students' mathematical ideas and guiding them through targeted questions. Student teachers report that interactions with the AI agents feel more authentic and insightful than standard preparatory exercises. At least 15 universities worldwide, including Stanford, the University of Helsinki, and ETH Zurich, have already tested or adopted similar AI-based systems for maths teacher training. Dr. Lee plans to release the agents and supporting resources publicly after further enhancements, aiming for broader use in teacher education.

This initiative responds to growing demand for scalable, tech-driven teacher training solutions. The AI agents are designed to improve how educators engage with varied student reasoning. If successful, the system could help address long-standing challenges in US mathematics education.