OpenAI Just Solved 370 Major Math Problems | Scott Aaronson artwork

OpenAI Just Solved 370 Major Math Problems | Scott Aaronson

MTS

October 17, 2026

Scott Aaronson, theoretical computer scientist at UT Austin, discusses OpenAI’s massive surprise drop solving 370 open math problems, including the quasi-Riemann hypothesis and new fast multiplication bounds.

0:00

Scott Aaronson

The Riemann hypothesis, many mathematicians would call that the greatest open problem in all of math.

But what we don't know is how accurate of an estimate that is. Now, in the Riemann hypothesis says, it actually is a very accurate estimate. David Hilbert was asked, I think, what would he ask if he were woken up 500 years in the future? He said, first thing I'll ask, has the Riemann hypothesis been proved? It still hasn't been, but a lot of the consequences of it, actually we now know from one of the results that OpenAI proved a week ago, which was called the quasi-Riemann hypothesis.

0:44

Theo Jaffe

And we're back, we're live with Scott Aaronson, who is a theoretical computer scientist at UT Austin. He used to work at OpenAI, and he writes the blog Shtetl Optimize, which is one of my favorites. I've been a reader for quite some time.

And he's also a former guest of the Theo Jaffe Podcast from like three years ago. So welcome back.

1:03

Scott Aaronson

Thanks, it's great to be back, Theo.

1:05

Theo Jaffe

All right, so what should we start with? Is there anything new happening in math?

So yeah, I guess we could start with broad takes on the whole situation. I read your post about the age of wonders and terrors almost a month ago. How are you feeling now?

1:25

Scott Aaronson

Well, I mean, we are still in a mode of responding to what is happening. I mean, my wife is at Berkley right now for a theoretical computer science program, which was supposed to be about other things, de-randomization. It is now about this. I think all across the mathematical research world, you have seen people drop what they are doing. You've seen them try to understand these solutions to many of their major open problems that OpenAI just dumped onto GitHub a week ago. This is the topic of conversation. So it would be fair to say that I've never seen a moment like this in real life in mathematical research. Some of my favorite problems were solved in there. And I've seen that even just within the last week, people have made a lot of progress on understanding some of these results. And so for me, a question is, do I throw myself into trying to understand these things, or do I wait for younger people to understand them, to re-explain them more comprehensively, and then try to understand what's going on?

It might be more of the latter for me. But a question that I posed 20 years ago with Greg Cooperberg, which was called the Unitary Synthesis Problem, was one of the 370 problems that OpenAI announced solutions to last week. And so just yesterday, there was a workshop at Columbia already, trying to explain and understand this proof. I joined that by Zoom when I was able to.

For all the anxieties that people feel about the future of the field, about what is there going to be left for us to do, it has given me some optimism, I think, to see people instead of just resigning or hiding under a rock somewhere, they are racing to try to understand this stuff. And it's like you may get a misleading impression if you're looking on online and just seeing people arguing about it. But I think actual experts are mostly just rushing to understand the new theorems that have just dropped on us.

4:26

Sophia

Yeah. So I wanted to ask you about this open letter of sorts that was published called A Severe Misalignment of AI in Mathematics. And I think the general-

4:36

Scott Aaronson

Was this the Association for Human Mathematics thing? Or was that a different one?

4:40

Sophia

It's math and ai.org.

4:42

Scott Aaronson

Okay. Oh, okay. Yes.

4:44

Sophia

So I guess the core premise here is like, there was a certain community effect of mathematicians getting together, solving problems conceptually, like discussing the conceptual leap whenever a problem was solved. This whole microcosm of humanity that they call it. And now there's this rush in the way that new solutions are announced. And I guess it takes away from the thoughtfulness and community of mathematicians trying to understand. So what do you think are the biggest risks of alignment between AI and mathematics? Is it really just that we're not fully understanding and metabolizing the conceptual leaps as a field?

What do you see?

5:29

Scott Aaronson

Yeah, okay. So I think that it would be crazy to not be worried right now about the future of the mathematical community, to whatever extent someone cared about that community, or was part of it, or had hoped to join it. I mean, this is an asteroid has landed on our world, right? This is not something that, people are in different stages of denial, grief, bargaining, acceptance, right? But I think a denial is just not going to be a viable strategy. This is not going away. And in fact, the model that OpenAI used is something that will just plausibly be available to regular subscribers a few months from now, right? So anyone will be able to get results like this. And I would say that since the summer, people have woken up to this. Our whole way of doing research is different now, right? Almost any paper that you read will contain near the end a little AI statement which explains which parts of the paper did they think of, which parts came from Ashtra or from Claude. So that is now just almost like without any fanfare just within the last few months, that became part of writing a paper, part of doing research.

Read the other 24 minutes of this episode here.

$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.

Not what you expected? Full refund within 14 days.

Thousands of transcripts fetched by people building searchable podcast archives

Fetch transcripts with code

The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:

request
curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.

Using your own key:

request
curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000796808325