Matt Green: verifying flashy model-generated math results is now the hard part
matthew_d_green · x · 2026-07-29
Matt Green argues that the main bottleneck for flashy new math results is no longer finding a counterexample or machine-checkable proof — it is verifying that the result is actually correct.
He says current models can happily generate convincing but false “result slop,” and unraveling it can take experts hours. For harder non-practical cryptanalysis, that verification problem is the real blocker.
Related event: Experts Highlight Challenges in Verifying AI-Generated Results(2 posts)→
More from AGI Musings
- “Pacing the Frontier” thread argues AI progress may need coordination mechanisms — TheZvi · 2026-07-29
- Google research on 15 million Gemini interactions finds little evidence of mass job automation — ChuckDBrooks · 2026-07-29
- AI is turning data pulls, forecasts, and A/B tests into one-line reports — yangyi · 2026-07-29
- AI now creates a much wider gap between average and power users — bendee983 · 2026-07-29
- Cambridge researcher says the incident is a warning shot about rising AI capability — S_OhEigeartaigh · 2026-07-29
- AI 2027 tracker says capabilities are still moving slower than the original scenario — shadowt1tan · 2026-07-29