Since generative AI arrived, most of higher education has reached for a policy document or a detection tool before asking the prior question: what is this tool actually for? Bearman et al. (2023) found that our own literature rarely specifies what it means by ‘AI’, which leaves the tool to define the problem it is meant to solve. I think that is the wrong way round, and this note argues for starting somewhere else.
Once you have named what a student must do for themselves, teacher presence turns out to be a variable you have to design in, not a given that survives a chatbot’s arrival on its own; Li et al. (2025) found engagement rose markedly when a visible teacher stayed present alongside the tool. From there, the more useful move is neither banning AI nor waving it through, but building critical evaluation of its output into the taught content itself, as McPhee and Jerowsky (2025) and Tran (2025) both argue from different directions.
A worked scenario, Elena Castellanos rebuilding a writing assignment around what she actually valued, shows what this looks like once the tool stops setting the terms.
Reviews
There are no reviews yet.