PME-NA

AI-Generated Feedback in Mathematics Education: Strengths, Limitations, and the Role of Teachers

• In Event: AI in Mathematics Education with Teachers

Description

This study explores the strengths and weaknesses of AI-generated feedback in mathematics education, specifically examining ChatGPT's instructional responses. While AI models offer accurate procedural guidance, they struggle to foster conceptual understanding and deeper student inquiry. The analysis of ChatGPT feedback highlighted common elements such as generic praise, direct instruction, and suggestions for visual aids, but often lacked scaffolding for critical thinking. AI feedback remains largely one-directional, focusing on correctness rather than diagnosing misconceptions or promoting reflection. The findings emphasize that while AI can support teachers, it cannot replace their ability to adapt feedback, interpret student reasoning, or cultivate meaningful discussions.

Authors

• Cindy Yovanov, Drexel University

• Jason Silverman, Drexel University

• Amanda Reinsburrow, Drexel University