# AI 'solved' famous maths problems. Read the asterisks.

> AI advanced real maths in May 2026 — but humans and Lean verified every step.

*The results are real. So are the caveats nobody screenshots.*

By The InsidersFeed Desk · InsidersFeed
Canonical: https://insidersfeed.com/news/ai-maths-erdos-read-the-asterisks

> **Key:** **The take:** something real happened in May — AI helped crack maths problems that sat open for decades. Now ignore the 'AI is smarter than mathematicians' takes, because that is not what the papers say.

Two claims, one week. OpenAI: an internal reasoning model disproved Erdős's 1946 unit-distance conjecture. DeepMind: **AlphaProof Nexus** solved nine open Erdős problems plus 44 conjectures. The dunk-tweet version — 'Google beat OpenAI nine to one' — is fun and almost entirely beside the point.

## Where the asterisks live

OpenAI's proof is human-checked and **not yet through formal peer review**. DeepMind's nine are machine-verified in **Lean** — which is the genuinely impressive part, but it also means the win comes from a tight generate-then-verify loop, not a model free-soloing genius. DeepMind even noted a basic agent solved the same nine; the fancier system mostly made it cheaper.

> **Note:** **Credit where due:** producing a novel construction from algebraic number theory, or a Lean-certified proof of a 56-year-old problem, is not nothing. This is real research output. Our gripe is with the framing, not the math.

The tell is that the people closest to it are the most careful: Hassabis said straight out the system is 'still not AGI'. When the vendor undersells and the timeline oversells, trust the vendor. AI is now a real instrument for parts of maths — narrow parts, with a human or a checker holding the pen.

## FAQ

### So is AI better at maths than humans now?
No. It cracked specific long-open problems with heavy human or formal-checker involvement. DeepMind's CEO explicitly called it 'not AGI', and OpenAI's result still needs peer review. It is a powerful tool for parts of maths, not a replacement for mathematicians.

### Who actually 'won', OpenAI or Google?
Wrong question. OpenAI disproved one famous, hard conjecture; DeepMind solved nine others by a different, machine-verified method. They are not the same task, so the 'nine to one' scoreline is a headline, not a result.