Showing and doing: ChatGPT and conversation analysis

Generative artificial intelligence programmes such as ChatGPT are the subject of much debate in higher education. The incorporation of these programmes into pedagogic practice has been hailed as significant in aiding students’ academic work through providing suggested conceptualizations, research approaches, and analyses. While these large language models can generate apparently human-like content, this article argues that, for students engaging with conversation analysis, the use of ChatGPT is problematic. It is argued that this is because this kind of AI is not rooted in the world of human language use where talk is a vehicle for accomplishing actions. This article presents a study that adopts a ‘thing’ ethnography to understanding the way that ChatGPT conducts conversation analysis. The study explores the differences between ChatGPT’s approach to this kind of analysis and myself as a researcher in this field. ChatGPT was also invited to offer a comparison of the two approaches as well as commenting on the value of the exercise for students studying this area in higher education.