OpenAI
Can AI actually do maths now? Explained
AI solved problems that stumped people for decades. Here's what that really means, simply.
Yes — in 2026 AI solved real, decades-old maths problems, checked by special software.
You may have seen headlines that AI 'solved famous maths problems' in 2026. Is that real? Mostly yes — here's what it means, in plain English.
How can we trust it?
This is the clever part. AI can sound confident and still be wrong, so researchers pair it with a 'proof checker' — software (called Lean) that won't accept a proof unless every single step is logically correct. So it's not just 'the AI says so'; a strict program has verified the maths actually holds up.
Does it mean AI is smarter than people?
No — and the experts are clear about this. The AI solved specific hard problems with human guidance and strict checking. The people who built it say it's 'not AGI' (not a general human-level intelligence). Think of it as an extremely powerful calculator for certain kinds of reasoning, not a replacement for mathematicians.
Sources
- An OpenAI model has disproved a central conjecture in discrete geometry — OpenAI, 20 May 2026
- OpenAI's milestone math breakthrough played to AI's strengths — Understanding AI, 22 May 2026
- Google DeepMind's AlphaProof Nexus solves 9 Erdős problems and 44 conjectures — Crypto Briefing, 26 May 2026