Abstract

This article presents a classroom activity using ChatGPT to support K-12 students in refining mathematical definitions. Definitions are central to mathematical understanding, yet students often struggle with the formal language surrounding them. To address this, the authors designed a prompt-structured task where students iteratively refined self-generated definitions through conversational feedback with ChatGPT. Two strategies, questioning and feedback, guided students to reflect on and revise their definitions across multiple rounds. Findings suggest that interacting with the LLM helped students articulate and improve their concept images, encouraged precise mathematical thinking, and provided timely, low-stakes feedback. Students responded variably to the task structure, revealing both benefits and limitations, such as technical access issues and prompt fatigue. Despite challenges, this pilot trial demonstrates that LLMs can serve as effective tools for promoting conceptual understanding and critical reasoning in mathematics. The authors advocate for continued exploration of AI tools that position students as active participants in their own learning. 

Keywords

Mathematical Definitions, Large Language Models (LLMs), Artificial Intelligence in Education, Concept Image and Concept Definition, Formative Feedback, Pre-service Teacher Education, Critical Reasoning], Large Language Models, (LLMs), Critical Reasoning

How to Cite