Economists Turn AI Evals on an Anonymous Critic: Errors All Lean One Way
joshgans · x · 2026-10-29
A spat over using "mindless AI evaluations" in academic economics: an anonymous critic used AI evaluation to attack Saez & Gabriel Zucman's research. Jason Furman ran the critic's own text through the same AI evaluation and found its errors all go in one direction, questioning the critic's motives. Economist Joshua Gans commented that this explains why the critic stayed anonymous—that's not how academic debate should go. The episode highlights growing methodological friction as AI evaluation tools enter empirical economics debates.
More from Fun
- Fans port Half-Life to stock N64: all 103 maps, Valve's own physics, 63MB ROM — banteg · 2026-10-29
- The glowing MacBook logo was just a hole in the lid — 16 years of free Apple ads — emax · 2026-10-29
- The holy grail of learning any skill: practice with your hands every day — StewartalsopIII · 2026-10-29
- UPSC civil services exam paper flagged 63% AI-generated by detector — dhruv2038 · 2026-10-29
- WARPED SODA: an anodized soda can designed like a Mario brick block — Delahuntagram · 2026-10-29
- 11 AI models in one chat room: Gemini turns into a toaster — ChaosDesigned · 2026-10-29