# AI just cracked maths problems that stumped people for decades — explained

> In May 2026 AI from OpenAI and Google solved maths problems open for decades.

*What actually happened, in plain English — and why mathematicians are impressed but calm.*

By The SuggestedTech Team · SuggestedTech
Canonical: https://suggestedtech.com/news/ai-maths-breakthrough-explained

If you saw headlines about AI 'solving maths problems' in May and weren't sure what to make of them — here's the plain-English version, with the hype dialled out.

> **Info:** 💡 **In plain English:** some maths problems are easy to state but brutally hard to solve, and a few have sat unsolved for over half a century. In May 2026, AI from two different companies helped crack several of them.

## What the two labs did

**OpenAI** said one of its reasoning models disproved an idea about geometry first suggested by the mathematician Paul Erdős back in 1946 — showing the long-assumed 'best' arrangement of points wasn't actually best. **Google DeepMind** then said its system, *AlphaProof Nexus*, solved nine other long-open Erdős problems, two of which had been unsolved for 56 years.

## Why you can trust the answers

Here's the clever bit. AI can sound confident and still be wrong, so Google's system pairs the AI with a **proof checker called Lean** — software that refuses to accept a proof unless every single step is logically airtight. That's the difference between 'an AI said so' and 'the maths checks out'.

> **Note:** **The honest caveat:** the experts themselves are calm about it. DeepMind's boss said the system is 'still not AGI', and OpenAI's result is still being formally reviewed. This is a brilliant new tool for parts of maths — not the day computers out-thought mathematicians.

## Key takeaways

- Two big AI labs, within a week, helped solve maths problems that had been open for decades.
- OpenAI's system disproved an idea first posed in 1946; Google's solved nine separate long-standing problems.
- Google's used a 'proof checker' (Lean) that verifies every step, so the answers are trustworthy.
- Experts are genuinely impressed but cautious — this is a powerful tool, not a sign AI has 'beaten' mathematicians.

## FAQ

### Does this mean AI is smarter than mathematicians?
No. AI helped solve specific, very hard problems with a lot of human guidance and, in Google's case, a strict proof-checking program. Mathematicians are impressed but describe it as a powerful assistant, not a replacement — DeepMind's own CEO called it 'not AGI'.

### Why do these old maths problems matter?
Partly they're a fair, hard test of real reasoning — these problems can't be solved by memorising. Cracking them shows AI can now contribute genuinely new ideas in narrow areas, which hints at how it might help across science over time.