Text-generative AI (GenAI) tools have become somewhat omnipresent in academia, even if researchers do not use them, they will surely have a story to relate or their own opinions on (in)appropriate GenAI uses. In a quantitative survey of researchers at Danish universities, Andersen et al. (2025: 2) found that researchers used GenAI to save time, be more efficient, to reduce stress and for reasons of self-esteem. Yet many academics are concerned about their peers’ negative evaluations of their own GenAI use (Andersen et al. 2025) and issues such as attribution and originality in written texts (Wu, Lu & Lin 2025). However, among the newly emerging literature on GenAI and its use by academics, there is to date little focus on actual documented practices.

In this seminar, I will present data from a linguistic ethnographic case study from the AI-UNI project, focusing on two science researchers collaborating on writing a chapter. The data constitute a research diary, based on a day-in-the-life approach (Tagg & Lyons 2021), participant made screen recordings and interviews. The data show how the researchers negotiate what a book chapter should be like, ask
ChatGPT if there is something missing from the content and discuss how ChatGPT can do tasks like writing an abstract “way better than” they could. Different disciplines value academic writing genres in different ways (Tusting et al. 2019) and these preliminary analyses point towards how the participant researchers conceive of GenAI as merely a tool to help them write their book chapter, especially when writing for them is not the main part of research in science.