Sampling is not understanding: why AI can produce proofs but not replace mathematicians
gerardsans · x · 2026-08-26
In a discussion thread, gerardsans argues that current frontier AI doing math resembles AlphaGo's move 37 — it performs computation and sampling, generating many proofs from which a few can be verified, yet possibly none are actually understood. Sampling is not truth and does not produce understanding.
He therefore contends AI cannot replace mathematicians or related jobs, lacking the judgment and lived experience true understanding requires; practitioners should use AI to their advantage, and if a proof is correct, celebrate it — science values understanding, not ego or status.
Related event: 'Sampling Is Not Understanding': AI Computes but Doesn't Grasp Math(2 posts)→
More from AGI Musings
- False Collatz Proof via a Lean Bug: Talk Probes Proof Assistants as LLM-Found Bug Targets — marc_lelarge · 2026-10-16
- Pennsylvania residents reject $10,000-per-household offer for massive Amazon data center — Polymarket · 2026-10-16
- A 1968 newspaper predicted today's AI art debate almost word for word — mtizard · 2026-10-16
- Curing AI-era FOMO: define what 'fulfilling your potential' means to you — dejavucoder · 2026-10-16
- "An LLM doesn't understand anything" gets the perfect parody reversal: neither do you — VoidStateKate · 2026-10-16
- Nina Schick: Europe's Policy Circle Still Doesn't Believe FSD, Orbital Compute Are Real — NinaDSchick · 2026-10-16