In the Turkey high-school experiment, students handed plain GPT-4 for homework scored 17% worse on the final exam than students without it, and believed they had learned more. The model's default is to hand over the answer, skipping the effort that creates learning.
A World Bank randomized trial in Nigeria paired a GPT-4 tutor with teacher guidance in a six-week after-school program and measured more than twice the effect of the most effective known education interventions, at very low cost.
AI beats most individuals at ideation. GPT-4 topped 91% of humans on a variant of the Alternative Uses Test, and Wharton judges preferred its product ideas to students'. But it produces one person's range, circling back to VR, blockchain, environment, and AI.
The ideation paradox: what you want is a diverse pool to choose from, and AI gives you variations on a theme. Worse, once you have seen its suggestions, anchoring makes it harder to reach ideas outside them.
Terms in this piece · Glossary
grounding — Tying a model's answers to checkable sources — retrieved documents, live data, tool results — instead of letting it answer from memory alone.
Why it matters
A research-groundingTying a model's answers to checkable sources — retrieved documents, live data, tool results — instead of letting it answer from memory alone.Full definition → rebuttal to the 'AI rots your brain' narrative, with concrete frameworks ('think first, write first, meet first') and shareable prompts for tutoring, ideation, and meeting facilitation — useful for anyone shaping how they and their teams actually use AI day to day.
Key quotes
“Ultimately, it is how you use AI , rather than use of AI at all, that determines whether it helps or hurts your brain when learning.”
“AI doesn't damage our brains, but unthinking use can damage our thinking.”
“Your brain is safe. Your thinking, however, is up to you.”
