GPT Solves Years-Old Gradient Descent Math Proof Baffling Researchers
konstmish · x · 2026-08-03
ML researcher Konstantin Mishchenko shared how an LLM helped crack a theoretical math problem that had bothered him since 2019. In their paper on adaptive gradient descent, the authors couldn't determine if a constant factor in their algorithm could be tightened to exactly 1.
After prompting GPT, the model concluded that the method would not converge even with a constant factor of 0.98 and constructed a PEP-style (Performance Estimation Problem) proof. This highlights the potential of frontier LLMs in handling complex, highly specialized mathematical reasoning.
Related event: Researcher Uses GPT to Solve Long-Standing Gradient Descent Proof(2 posts)→
More from Fun
- Dev vibe-codes a living-room wall game with Codex, controlled by hand gestures — tristanbob · 2026-09-22
- Siri can't even make phone calls anymore, it just stares at you — ikeadrift · 2026-09-22
- TIL: Jev is named after Jevons Paradox, not the encephalitis virus — tristanbob · 2026-09-22
- Grok prompt turns your avatar into a stitched circular patch design — bennash · 2026-09-22
- ~20% of employees at every AI company now follow the same X account — signulll · 2026-09-22
- Chang Robotics founder loses bet, does factory acceptance test in Buffalo — MatthewChang · 2026-09-22