// NATURE NEWS — SPAZIO & SCIENZA
Universities must measure learning, both with and without AI assistance
Search author on:
PubMed
Google Scholar
Marc Schmitt argues that the danger of artificial intelligence lies in passive use, and that universities should teach students to argue with AI models and assess how well they do so (see M. Schmitt Nature 657, 575; 2026). One measurement is missing. Assessing how a student spars while the model is present captures performance with the partner in the ring, not what the student can do once the partner leaves. The two can diverge: in a randomized trial involving secondary-school mathematics students, unrestricted chatbot access raised practice scores, but lowered performance on a subsequent exam taken without it (H. Bastani et al. Proc. Natl Acad. Sci. USA 122, e2422633122; 2025). An evaluation that stopped at the assisted session would have called that tutor a success.
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
Prices may be subject to local taxes which are calculated during checkout
Chinese doctoral students in clinical medicine can graduate without a thesis — more degrees to follow
In Indonesia, PhD students blow the whistle on bad supervision
Will AI grant reviews make science more conservative?
Meet Neuropixels: how probes that track hundreds of brain cells at once are revolutionizing neuroscience
Patrick Stewart dies: how Captain Picard inspired today’s science leaders