Discussion

AI-assisted team proves a 30-year-old mathematical conjecture, CGTN reports

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#8021

A Chinese research team says it has proved the 3D Ghys conjecture, a problem that had resisted mathematicians for more than 30 years. The work is a striking example of AI helping with advanced mathematical reasoning, though researchers guided the effort and checked the resulting proof.

Watch Desk analysis

What happened

CGTN reports that a team led by Xu Disheng, a professor at Greater Bay University and visiting scholar at Shanghai AI Lab, worked with the lab’s Intern-S2 science model. The model handled large, complex chains of mathematical derivation; researchers steered the work and sifted promising lines of reasoning.

A separate team at Peking University then used its own framework and AI agents to formally verify the core theorems. CGTN says that process generated hundreds of thousands of lines of verification code. The paper is available as a preprint, and the verification code has been open-sourced, according to the report.

Why it matters

The result is more than a model producing a plausible-looking proof. The reported workflow combines AI-assisted exploration with formal checking, an important distinction when mathematics demands that every logical step hold. The conjecture’s real-number counterpart remains open, CGTN notes, so this is a significant result, not a universal maths-solving machine arriving in a lab coat.

Our read

The useful question is not whether AI has replaced mathematicians. On this account, it has not: people set direction and judge which reasoning to pursue, while separate researchers checked the proof formally. But the collaboration suggests models may help researchers explore difficult problems at a scale or pace that would otherwise be hard to manage. The preprint and open verification code give other mathematicians something concrete to examine.

What to watch

  • Whether the preprint and verification code stand up to scrutiny by other mathematicians.
  • How much of the proof came from the model’s derivations and how much from researchers’ guidance.
  • Whether the approach proves useful on other difficult mathematical problems.

Discussion spark: If AI helps produce a proof but mathematicians guide the search and formally verify the result, how much credit should the model receive?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.