이화여자대학교

초록/요약

This study analyzed the argumentative thinking processes and experiences of university students collaborating with ChatGPT in ill-structured problem solving. For this purpose, ChatGPT conversation scripts from 18 consenting students were analyzed, and in-depth interviews were conducted with 12 participants discussing their interactions with ChatGPT. In ill-structured problem solving, clear logical reasoning and judgment are essential for argumentation. Therefore, the argumentative thinking process was considered in developing a tool to analyze ChatGPT conversation scripts. Analysis of 1,095 learner utterances to ChatGPT revealed 756 inquiries for internal argument structure, 217 for argumentative writing, and 122 for other purposes. An analysis of the patterns of learners’ inquiries over time showed that queries about ‘reasons,’ ‘claims,’ and ‘evidence’ comprised a high proportion throughout the interaction. In the early stages, there were many utterances about ‘content organization’ and ‘feedback,’ while in the later stages, there were numerous utterances concerning ‘body writing’ and ‘writing format’ for report preparation. In terms of the complexity of internal argument structure, most learners tended to make utterances using only three elements (claims, reasons, evidence). Furthermore, through in-depth interviews, the unique characteristics of the interaction experiences between learners and ChatGPT were analyzed and presented. This study holds educational significance in educational technology as it views ChatGPT as a subject interacting with learners and proposes a new instructional method for ill-structured problem solving utilizing this interaction.