Mathematicians debate whether LLMs are filling the 'convex hull' of provable theorems
notmisha · x · 2026-11-06
Noah Chrein argues that LLMs, backed by the world's compute, are effectively filling in the 'convex hull' of standard mathematics — combinations of existing theorems and objectives that could have been proven but never were. What remains lies outside that hull, which massively favors trained mathematicians who know when new math must be invented versus pieced together.
The replies devolved into mathematicians arguing whether the correct abstraction is a convex hull, a semigroup, or a linear span — which the poster calls the most mathematician-possible response.
More from AGI Musings
- Anthropic report: full AI automation could yield 15-16x research output, labs remain the bottleneck — teortaxesTex · 2026-11-06
- In 2026 AI agents still talk to each other in human language about human programming languages — mark_k · 2026-11-06
- Oxford Position Paper: From Autonomous Agents to Cognitive Symbionts — Oxford · 2026-11-06
- Yacine: everyone is doing large-model RL wrong, long-horizon rollouts insane — yacineMTB · 2026-11-06
- "Give it just a few years": AGI framed as humanity's next alien-invasion moment — AndyMasley · 2026-11-06
- CDT+GT and indexical uncertainty in multi-agent clone settings: one policy network or many matters — jessi_cata · 2026-11-06