Harvard's Gary King: AI reviews beat top human peer reviews on every measure
anderssandberg · x · 2026-10-29
A 2026 paper by Gary King and Steven Worthington, "AI Article Reviews Dominate Human Peer Reviews at Their Best", shows AI-generated article reviews outperform reviews from well-compensated human experts on every measure the authors could construct — even reviews judged unusually high-quality by editors and authors. The paper explores policy implications: concerning downsides, but also paths to faster, higher-quality publishing, fewer adversarial interactions, and better experiences for editors, reviewers, authors and readers. Anders Sandberg adds that the interesting question is turning this cheap-but-imperfect reviewing machine into something genuinely truth-tracking, noting you can opt out of AI but others will use it — sometimes maliciously.
Related event: Harvard Study Finds AI Paper Reviews Outperform Human Peer Reviews(2 posts)→
More from AGI Musings
- A 4-person team ships 77,000 changes/month with AI agents — how devs stay relevant — yoobinray · 2026-10-30
- Anthropic API Key Leaked; Agents Detected and Rotated It Within 15 Minutes — tylerjdunn · 2026-10-30
- Alternate 1865: Steam-Powered AI Agents and Britain's 'Red Tie Act' Output Cap — azmackay · 2026-10-30
- Dev quips that every AI headline you see is already six months old — 0xsachi · 2026-10-30
- AI x Bio criticized for avoiding scholarship; 12-person Leash Bio ships 3 papers — niloofar_mire · 2026-10-30
- Medicine's labor shortage will be huge, and medtech won't fix it — PAstynome · 2026-10-30