AI audit finds 2,035 errors across 138 Saez-Zucman economics papers
lemire · x · 2026-10-29
Tyler Cowen published an AI audit of economists Emmanuel Saez and Gabriel Zucman's research, and Lemire notes anyone can now take apart a scholar's work.
- 138 papers audited, 2,035 problems found; 147 serious enough to change a headline number; 69% of papers had at least one
- 97% of directional major/moderate errors overstate the paper's claim
Method: two independent audits per paper (Claude Opus 5.5 and OpenAI GPT-6 Astra, blind to each other) recomputed headline numbers from the authors' own files and original sources; a third adjudicator agent re-verified, tried to refute, and graded each finding.
Cowen calls for a radical revaluation of this work and its policy arguments, predicting AI will revise economic science with status losses often outweighing gains.
Related event: AI Audit Finds 2,035 Errors Across 138 Economics Papers(4 posts)→
More from AGI Musings
- Google researcher: people posting these AI-work demo videos are 'doing nothing of consequence' — prajdabre · 2026-10-29
- AIxBio builder: people build the models they can, not the models they should — hardimanjames · 2026-10-29
- BallotBench pits 9 frontier AI agents against experts in live midterm forecasting — karinanguyen · 2026-10-29
- Pedro Domingos: OpenAI and Anthropic are the WorldCom and Global Crossing of AI — pmddomingos · 2026-10-29
- Opus videos that wowed two weeks ago already look like slop — our hunger is infinite — jzlegion · 2026-10-29
- Burny: AI progress is more continuous than most assume — it's continuous hillclimbing — burny_tech · 2026-10-29