Gary Marcus rebuts Scott Aaronson's singularity claim: benchmark curves show no monthly capability explosion
GaryMarcus · x · 2026-09-17
Gary Marcus amplified Nate Witkin's critique of Scott Aaronson's essay arguing the singularity has already started, calling it "unmoored from reality." Witkin cites 1 year of benchmark scores on difficult knowledge-work tasks to argue capabilities are not "dramatically ramping up every month," and adds two caveats: (1) these are pass@1 scores, while extrapolating to real-world "drop-in worker" automation requires pass^n — succeeding n times in a row; (2) even strong benchmarks lack external validity, as they cannot model field-specific regulatory, security, and organizational constraints of real knowledge work.
More from AGI Musings
- Blogger claims the next generation won't need a college degree to live well — rand_longevity · 2026-11-02
- Carl Frey: 85% in China see AI as beneficial vs 38% in US — resistance maps to expertise costs — carlbfrey · 2026-11-02
- After a painful rebuild, fejau concedes: build on Hermes and you own your AI for life — Teknium · 2026-11-02
- 'Meatwashing': manually rewriting AI output to hide its origin, and why it's a double waste — JeremyNguyenPhD · 2026-11-02
- AI safety researcher: AI companies entirely lack this kind of effective human oversight — davidmanheim · 2026-11-02
- Long-read: AI's real value comes after the efficiency era—infra glut, agent tooling boom — yangyi · 2026-11-02