Understanding the Impact of Generative AI on Professional Judgement in Accounting

This paper explores how generative artificial intelligence (GenAI) reshapes the exercise of professional judgement in accounting, drawing on ethnographic fieldwork and semi-structured interviews conducted within a mid-tier UK accounting firm. We examine how accountants engage with large language models (LLMs), particularly ChatGPT, in their daily work. The study adopts a sociomaterial genealogical lens (Scott & Orlikowski, 2025) to conceptualize GenAI not as a fixed technology but as a phenomenon in-the-making, whose influence unfolds through ongoing human–machine interactions. Building on this, we develop a new theoretical framework (illustrated in the accompanying figure) that maps four emergent modes of interaction — Human-led, AI-moderated, Encroachment, and Non-use — along two axes: the explicit/tacit nature of knowledge and the criticality/instrumentality of use.