OpenAI says its AI solved major maths problems. Experts are still checking
What happened: On 26 October 2026, OpenAI said its AI models had made three new discoveries:
- Maths: a proof of the local smoothing conjecture in every dimension. It had only been proved in three dimensions before.
- Computer science: a claim that disproves the Exponential Time Hypothesis, a widely accepted idea about how long certain hard problems take to solve.
- Physics: a “mathematically rigorous” proof related to quantum chromodynamics, the theory of how the particles inside atoms hold together.
What experts said: Mathematician Alex Iosevich of the University of Rochester told Al Jazeera that OpenAI appeared to have “cracked” the conjecture, calling it a “historic development”. MIT’s Ryan Williams said disproving the Exponential Time Hypothesis would be major, but that researchers are still checking whether the proof holds up. OpenAI itself admitted that letting its chosen expert groups see the work first “creates asymmetry of information”.
What it means, in plain words: These are proofs, not products. Williams said the work is unlikely to have real-world uses soon. But if the proofs hold, AI has moved from helping researchers to producing new results of its own, which could speed up science over time.
The hype check:
- These are company claims until other experts finish checking them. Last month, a different AI company’s discovery claim got mixed reviews from scientists.
- The same AI that writes a new proof can still make simple mistakes in everyday chats. Always check important answers.
Want to understand how these models work? Start with How do chatbots like ChatGPT actually work?