Adopting Artificial Intelligence in Schools: Opportunities and Challenges in the Diffusion of a Commercial Large Language Model Across a Teaching Population

Artificial intelligence (AI) is the latest in a long line of digital technologies introduced into schools with promises of transformation and disruption. This paper examines the introduction of Microsoft Copilot, an AI productivity tool and commercial Large Language Model (LLM), in an independent secondary school in Australia. Drawing on in-depth interviews with six teachers across multiple faculties and teacher logs of user experience, the study explores how Copilot moved, or, in many cases, failed to move, through a teaching population. Findings reveal a contradictory landscape. Copilot was embraced when it offered clear utility, generating classroom resources, drafting communications, or producing creative outputs, yet faltered when teachers encountered opaque licensing restrictions, unreliable functionality, generic professional learning, and what several described as “AI fatigue.” Situated within diffusion of innovations theory and critical scholarship on educational technology, the paper argues that AI adoption in schools is best understood not as a neutral process of innovation uptake but as a contested social practice. Teachers’ responses to Copilot highlight the importance of transparency, subject-specific professional learning, and respect for professional expertise. More broadly, the case raises questions about the managerial and corporate logics embedded in contemporary AI, and their compatibility with the values and practices of schooling.