Fixing teachers’ problems? exploring teachers’ repair and maintenance work around generative AI technologies

The promise that AI tools will relieve teachers of tedious tasks and liberate time for more ‘valuable’ work with students has become a dominant narrative. However, the implementation of automated technologies raises questions around how human labour is implicated and situated in these processes. From the perspective of maintenance and repair studies we approach the often-hidden labour that teachers undertake when engaging with generative artificial intelligence (GenAI) technologies. Drawing on interviews with Swedish teachers we expose the tools’ inherent limitations in teaching contexts, and highlight the pedagogical, social, and moral dimensions of educational practice. We argue that these essentially inexplicable and ungeneralizable aspects of teaching – grounded in teachers’ situated awareness, professional experience, and tacit knowledge – cannot be codified or reduced to training data. Our results demonstrate the need for a radical rethinking of the forms of AI that education could benefit from – and those it should resist.