ChatGPT 기반 대화에서 나타난 물리 역학 개념 설명 구조와 오개념 수정 양상 분석

초록

This study analyzes ChatGPT-based dialogue data to examine conceptual explanation structures and misconception correction patterns in physics mechanics, and to explore their educational potential and limitations. Focusing on commonly misunderstood topics in mechanics—such as inertia, centripetal force, centrifugal force, and direction of motion—the study examines ChatGPT’s strategies for conceptual explanation, misconception detection and correction, use of analogies and examples, and feedback structures. The analysis reveals that ChatGPT can contribute meaningfully to students’ conceptual understanding by offering scientifically consistent explanations, repeated feedback, and analogies that promote visual imagination. In particular, ChatGPT responds to misconceptions with definition-based and step-by-step explanations, inducing cognitive conflict and facilitating conceptual reconstruction. However, limitations were also observed in ChatGPT’s inability to detect nonverbal cues, address complex problem-solving tasks, and provide emotional interaction. Accordingly, this study proposes that ChatGPT should not replace teachers but be utilized as a collaborative tool to enhance instructional strategies. The findings highlight the need for future research into AI-supported instructional models that integrate teacher–student–AI interactions in science education.

키워드