AI Skeptic Argues Lean Math Proof Success Comes From Search Algorithms, Not Intelligence
gerardsans · x · 2026-11-21
In a thread with @thinkingshivers, gerardsans argues that AI labs have misrepresented AI's role in recent Lean proof breakthroughs, attributing to 'intelligence' what is mostly search algorithms and software. His points:
- When a solution is well-defined and reachable via code, a search algorithm can find it
- Since AI lacks real understanding, it needs a Lean runtime to verify every step, can't explain why a result holds, and fails far more often than it succeeds
- He calls the intelligence explosion narrative false
- @thinkingshivers counters that the cope 'AI is great in verifiable domains but terrible in non-verifiable ones' has held up quite well so far
Related event: Debate flares over claim that AI only excels in verifiable domains(2 posts)→
More from AGI Musings
- AI autonomy benchmarks may be wrong, says Ramez; Mollick adds caveats — eldonredwards · 2026-11-21
- Counter to LeCun: LLMs May Skip Human-Level and Deliver Superhuman Intelligence — hbouammar · 2026-11-21
- Yoav Goldberg: many harness abilities shift from essential to marginally useful — yoavgo · 2026-11-21
- 'Frictional AI': Adding Resistance to AI to Prevent Human Skill Decay — JeremyNguyenPhD · 2026-11-21
- AI Is Testing What Economists Know About Bubbles — bloomberg · 2026-11-21
- Do executives owe any duty to future shareholders before AI's shoe drops? — AaronBergman18 · 2026-11-21