Content-type: text/html Downes.ca ~ Stephen's Web ~ Student use of LLMs can inhibit learning

Stephen Downes

Knowledge, Learning, Community

I link to this discussion on the Learning Engineering discussion list with a somewhat misleading title. It's based on a LinkedIn post from Ethan Mollick stating "AI can help learning... when it isn't a crutch." Mollock cites three papers which I reviewed for this post: first, AI Meets the Classroom: When Does ChatGPT Harm Learning? which states "Using LLMs as personal tutors by asking them for explanations improves learning outcomes whereas excessively asking LLMs to generate solutions impairs learning." Second, Generative AI Can Harm Learning says "students attempt to use GPT-4 as a 'crutch' during practice problem sessions, and when successful, perform worse on their own" though "These negative learning effects are largely mitigated by the safeguards included in GPT Tutor." Third, Effective and Scalable Math Support says "chat-based tutoring solutions leveraging AI could offer a cost-effective and operationally efficient approach to enhancing learning outcomes for millions of students globally." All these results are, at worst, mixed, and at best, show genuine promise in AI for improving learning.

Today: 5 Total: 87 [Direct link] [Share]


Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Copyright 2024
Last Updated: Oct 16, 2024 04:05 a.m.

Canadian Flag Creative Commons License.

Force:yes