AI-Assisted Proofs Claim Breakthroughs on Navier-Stokes and the Riemann Hypothesis

Unforced Navier-Stokes Singularity and the Generalized Riemann Hypothesis

AI-Assisted Proofs Claim Breakthroughs on Navier-Stokes and the Riemann Hypothesis

Terence Tao reports that Levent Alpöge, Tristan Buckmaster, and Matei Coiculescu, with heavy AI assistance, have produced a human-readable, Lean-formalized proof of finite-time blowup for the unforced incompressible Euler and Navier-Stokes equations in three dimensions. Alpöge also generated an enormous Lean-only proof of the Generalized Riemann Hypothesis using an internal AI model; verification and extraction of its strategy are still underway.

Levent Alpöge, using an internal AI model, has generated an (enormous) Lean-only proof of the Generalized Riemann Hypothesis.
  1. softwaredoug

    Something I don’t quite know, maybe someone closer to math can answer this question

    It seems we’re making tremendous progress on the back of existing abstractions and concepts with the brute force ability to generate language and verify proofs.

    For example, Tao says about the Riemann proof:

    > appears to use the full power of modern developments on the Langlands program and arithmetic geometry

    What I wonder: how do we create new abstractions and concepts? Like whatever “Langlands program” is? That then feedback into the system, and allow for even more sophisticated mathematics.

    Are we assuming AI will do this? Or does this still need to be something at least agreed-upon by a mathematics community - in the same way programmers talk about a “linked list” without having an hour long conversation to define it axiomatically? And without that social aspect, without breaking down long proofs into Lego block common abstractions, math may plateau? In part because we lose the ability to ask meaningful questions?

  2. PopePompus

    Could someone explain the difference (if any) between the Generalized Riemann Hypothesis and what has historically been called the Riemann Hypothesis?

  3. chunky1994

    https://leanprover.zulipchat.com/#narrow/channel/113488-gene...

    It looks like two independent kernels accept the result, which means its likely that there are no kernel bug exploitations in the proof.

  4. swingboy

    Dumb question but with something like 15 million lines of Lean code, how can we tell when the Lean build “passes” that it is actually formally verifying the right thing and there wasn’t just a “1+1=2” making it pass? Like you hear people say “if your Rust program compiles then it’s correct” but that never meant your business logic is correct. I’m not suggesting malfeasance or intentional error, just that maybe the model spit out something that “passes” but doesn’t actually prove what it says it does.

  5. actusual

    Is anyone else struggling with feelings of both excitement and terror? I haven't felt this much personal uncertainty about my career/life since I was an L1 MLE (10 years ago).

  6. gehsty

    Future of mathematics has two frontiers - one that LLMs can derive and another that we can understand. When the science studied is completely abject, what is the point of an LLM doing it.

  7. sega_sai

    As some others in the thread, I truly find this moment scary. I feel that right now AI essentially completely surpassed human intelligence. For now the results from AI are limited by what's easy to do within the computer "with pen and paper", but surely it's matter of time before this is unleashed in other fields.

    For a scientist this is scary, because it is unclear if there is space for research by people anymore. But I start to worry about actual dangers from this intelligence and where it can lead.

  8. maplet

    I understand that Tao does not want to promote a result which could turn out to be incorrect, but I simply cannot imagine posting about a Riemann Hypothesis result and not commenting about the significance.

More from this day

2026-11-20