Blog — Latest in Math Tech & AI
News, studies, deep-dives, and trends at the intersection of mathematics and AI.
Stellar Colosseum coordinates AI agents for longer mathematical proofs
A September 14 preprint studies how planning and criticism can organize AI mathematical research.
Fields medallists urge AI developers to prioritize mathematical understanding
A September 11 declaration challenges the use of solved problems as the main measure of mathematical AI progress.
AI-assisted combinatorics study reports an exact four-column formula
A September 10 preprint combines AI-assisted discovery with exact checks in a specialized extremal counting problem.
Geometry study pairs language models with symbolic deduction
A new preprint separates diagram reading from deduction to make geometry solutions easier to inspect.
OpenAI reports an AI-generated Navier–Stokes proof
A September 8 announcement releases a proposed proof and Lean code for scrutiny, with smooth forcing central to the claim.
Anthropic reports a complete Lean formalization of Fermat’s Last Theorem
The September 4 announcement puts AI-assisted verification of established mathematics in focus.
AutoGraphForge tests a full pipeline for AI-assisted graph theory discovery
A September 3, 2026 arXiv study links conjecture generation, counterexample search, novelty filtering, Lean formalization, and neural theorem proving.
ProofEvolve turns failed proof attempts into reusable theorem-proving progress instead of wasted search
Posted to arXiv on August 28, 2026, ProofEvolve presents a neuro-symbolic theorem-proving workflow that keeps verified partial proof structures and reuses them across future Lean problems.
MathAdv tests what theorem provers really know when mathematics stops looking uniform
Posted to arXiv on August 26, 2026, MathAdv introduces a benchmark designed to show where modern theorem provers transfer well across mathematical domains and where their apparent strength still depends on narrow coverage.
Showing page 5 of 10 · 86 total