Universities should assess learning with and without AI assistance
Marc Schmitt argues that the danger of artificial intelligence lies in passive use and that universities need to teach students to argue with AI models and evaluate their performance without the tool. He points to a randomized trial of secondary‑school mathematics students where unrestricted chatbot access raised practice scores but lowered performance on a later exam taken without the bot.
Key points
- Universities need to measure student learning both with AI assistance and after the AI is removed
- A randomized trial found chatbot access improved practice scores but reduced exam performance without the bot
- Marc Schmitt warns that passive AI use is risky and urges teaching students to argue with AI models
The study, cited from Proc. Natl Acad. Sci. USA, shows that measuring only the assisted session would label the chatbot as a success, while the post‑access exam reveals a decline in independent ability. Schmitt recommends that institutions adopt dual assessments—both during AI‑assisted work and after the AI is removed—to capture true learning outcomes.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- Alex Bores launches Who Decides to push AI safety into 2028 Democratic platform · 1 src
- CPS prosecutor cites imaginary case to high court after AI search · 1 src
- Anthropic’s watermark loophole lets users bypass marks on two older Claude models · 4 src
- Trump orders AI rebrand to 'Super Intelligence' in executive order · 1 src
- Anthropic and OpenAI set up election war rooms amid AI misinformation risks · 2 src
Comments
via GitHub Discussions