Skip to content
All tags

#learning-science

2 posts
learning deep-dive Learning How to Learn

AI Makes Things Smooth Exactly Where They Should Be Hard: What Generative AI Does to Learning

The single most-cited meta-analysis on ChatGPT in education (g = 0.867, ~500k views) was retracted by Nature in April 2026. But the positive finding was not overturned — the issue is that it measures performance while the AI is available. Bastani's PNAS RCT measured something else: +48% accuracy during practice with GPT-4, then 17% below never-users once access was removed.

#ai-and-learning#learning-science#llm#metacognition#self-learning
learning deep-dive Learning How to Learn

Learning How to Learn: Auditing the Course 4.17 Million People Took — What Holds Up, What's Just a Metaphor

Dunlosky's 2013 review rated 10 study techniques; only self-testing and distributed practice earned 'high utility'. But a 2026 systematic review puts the effect at 0.22–0.46, and Pan & Rickard's transfer meta-analysis finds 'no positive transfer' once publication bias is corrected — making the premise in the framework's own name the piece that tests worst.

#learning-science#self-learning#retrieval-practice#spaced-repetition#metacognition