AI와 인간의 토론 담화 분석 - AI 시대 토론 교육의 인간 중심적 재정립 방향 탐색 -

Abstract

This study aims to compare the discourse of debates between artificial intelligence (AI) and human speakers in order to explore differences in argument structures and persuasive strategies, and to derive educational implications. The data consist of two publicly available YouTube debate videos: one between a lawyer and AI (ChatGPT), and another between science and engineering university students and AI. The analysis, based on a discourse-analytic approach, focused on aspects such as amount of speech and control of the floor, pointing out logical inconsistencies, issue management, use of persuasive strategies, formation of collaborative arguments, and meaning reconstruction. The results showed that in the lawyer–AI debate, identification of logical contradictions, segmentation of issues, justification through analogies were prominent, with a notable shift from conflictual to cooperative discourse. In contrast, the student AI debate was dominated by arguments grounded in personal experience and emotional persuasion, while AI tended to secure logical and procedural superiority. These findings suggest that the dynamics of AI human debates vary depending on the participants’ expertise and the discourse context, and they highlight the potential of debates with AI as a tool for fostering critical thinking and argumentation training. However, this study is limited as a preliminary exploration confined to two cases, and future research should broaden the data and conduct experimental validation.