upvote
It's discomforting only if you attribute value to your ego and your pride. Nothing has changed for the mathematician that find intrinsic value in the field and are interested in understanding the proofs, or the nature of Maths itself.

It has only changed for the people who want to stamp their name in blood on proofs and then get social credit and clout for what they have done. In other words people who use the field to find extrinsic value.

This is really good for mathematics, and science in general.

reply
This misses the point I was trying to make, though I guess I didn't make it very clearly. I'm not concerned about who gets credit. I'm saddened by the lost opportunities for the wider mathematical community. There could have been conferences and Zulip chats and lunch conversations among young graduate students all making progress toward Riemann together with AI tools. It could have been a great problem to rally around. Hell maybe it could have been a Polymath Project. Instead it was solved using possibly tens of millions in compute behind closed doors.

But instead it will probably discourage many more mathematicians to leave math altogether, because they feel they can't compete.

reply
Is it actually good for mathematics and science? Do these things have some independent, objective value, independent of the people that interact with them?

If a new miracle drug is developed but only ever used by the person who developed it, would we applaud it as a scientific achievement?

You are dismissive of ego and pride, but those can also be components of what drive people to accomplish great things. If you take away the incentives to pursue subjects like this, what is left?

I’m actually not all that negative about the results, I believe these fields will adapt and we will move forward. Or at least that’s still a possible outcome. But I’m challenging your assertions here as stated.

reply
Wouldn't you say that a miracle drugs value is in curing people rather than in who gets a Nobel Prize from developing it? I can guarantee you that patients that have pancreatic cancer don’t give a shit if a human or a server rack developed a solution.

I would hence claim what is left is the people who are actually in this for the interest and understanding (intrinsic motivation).

reply
I’m making a subtler point about medical research. Yes, its value is in the benefit it brings to people. If someone developed a powerful new drug and didn’t let anyone use it, both the drug and its developer’s accomplishments would effectively be worthless. Similarly, if we produce proofs only machines understand, and in doing do deprive people of an opportunity to strive towards a goal, to figure something out for themselves — the value drops to near zero. Often in math, the problem itself is worth much more than a single solid.

I think even with intrinsic motivation, people would get worn down by having things they worked for, possibly for years, yanked out from under them on a regular basis. The field must change how it does work and provides incentives, or it will become no more than a hobbyist playground for LLMs, and largely stop providing benefits to humans.

reply
Very few if any human can hold the entire arch and workings of a modern CPU in their head either, yet people seem to be using their phones just fine and the value has not dropped to 0. If LLMs designed a chip (see HN thread from yesterday), and people don't really know every single in-and-out of it, but it works fine in the phone and gives you 1 month of battery lifetime, people wont care.

And again if having a problem solved before you "yanks it out from under you" that's still just wanting to be the one who crossed the finish line first. If you have internalized the extrinsic value mindset in this way, you will of course be sad about the developments.

Finally I think the current structure is already fine. Philosophy, classic literature, music theory departments still exist, and nobody is dropping out of them just because things were already done and figured out. I think here again we get to the archetype of the extrinsically motivated mathematicial who wants to acquire, stamp and claim things for their own for social status.

I think this will lead to people just doing maths in the way they want to, if I analyze Dostojevsky because I love it and find it meaningful, no-one can "yank" anything from underneath me and I can know that everyone of my peers have the right motivations to be there.

reply
Even if humans are never able to understand the proof, they can still go forward assuming the hypothesis is now true e.g. they can assume GRH when working on other problems.
reply
What's left is plenty enough of smart people. The whiny math subcommunity is not representative of all mathematicians or math-interested people. They are just very vocal.
reply
But people will always value ego and pride, even if we’ll live in post scarcity utopia there will be still status games and so on. And that’s good, it will keep things fun!
reply
Let's all applause the advancement of human society. I'm sure all Mathematicians agree.
reply
Hopefully it actually is advancement of human society, because thus far this and all the other breakthroughs seem to be the advancement of shareholder profits, with any consideration for the rest of humanity being secondary scraps falling off the table.

I really, really hope to be wrong with this cynical view. I don't know what we should do in the alternate case. One idea is to internationalise the computational resources for AI so it could be used for the greater good of humanity, but I guess that idea would get a bunch of resistance, so I dunno. I guess the last one shall turn off the lights.

reply
From what I see many people, for various of reasons, are OK to ditch the mass for that advancement. So the advancement might as well happen, just not for the common people.
reply
Many intrinsically driven ones do agree. We are getting answers and seeing discoveries at a pace that no-one could have imagined to see in their lifetime.

If you however are full of yourself, want to stand in front of YOUR peers and prove that YOU did this, YOU figured out how this is connected, you want to have YOUR name written on some gold coin or certificate hanging in YOUR office, YOU YOU YOU everywhere, then you will have a big problem with the current developments.

reply
A lot of this is the byproduct of how society views accomplishment, at least in the western world. While there are certainly lots of self centered, egotistical people out there, there are also many who are shaped by the constraints of their work and education. There aren’t many places where people get to the next rung by pointing out all the great things their teammates did. Ever since young school ages, we are taught to accomplish things on our own. Our grades are the byproduct of our effort, etc…
reply
This is a western capitalist/neoliberal view. In a lot of countries for example, family is more important than work. Having a child and a social circle is valued more than being a CEO.

In some middle-eastern countries, the value is tied to how good you can hold the room in a social setting.

Seems like the western world will have to adapt their values and education in the near future.

reply
Correct, which is why I specifically pointed that out from the get go.

I don’t see why the western world “will have to adapt”. This has been the case here for hundreds of years, and I don’t see a sudden shift to valuing different things. Maybe as generations go by, sure, but the west is in love with capitalism, the cars, the money, the “prestige”, it’s all quite sad.

reply
I totally agree with you, man, I totally agree.
reply
That many people want to leave their mark on the history of mathematics or science in some way, small or large, is not so narcissistic a goal as you are portraying it. The desire for one's discoveries to matter to the people around oneself is intrinsic and powerful, and without it, the passion of science gives way to mere pragmatism. We can both appreciate the new pace of discovery and mourn the damage that is being done to the spirit of human camraderie on the edge of discovery.
reply
If your goal is to impress other around you with your discoveries, it assumes that you are able to provide something of value that people around you find impressive.

This is extrinsic motivation. If this is your purpose, you can pivot and do commercial business and make a lot of money, or something else that people around you find meaningful. People might value how good communicator you are, how much empathy you have, how caring you are for your family, etc.

But then it wasn't about the math nor understanding.

Imagine reading a book for the sakes of the book. You have your imaginary world you get into, you will keep reading the book even if there are 1000% better ways to create value.

Now imagine you read the book because it gives you social importance or skills, you are able to speak about a topic to impress and acquire for example sex from others. That is a different motivation, similar to what many of these mathematicians have.

reply
You just describe 90% of Academia.
reply
That's an underestimate.
reply
Actually argument and I were talking about different things. He/She was talking about getting a name on a paper. I don't give a fuck about that (and I don't work in Academia anyway). I just want to keep my job.

But yeah I should applause thunderously for the advancement of human society, to be a good boy.

reply
Indeed! The greedy, selfish mathematicians taking all the theorems have been a problem in our society for far too long!

It's enraging to think that someone like Riemann would name a hypothesis all after himself. He was stamping his name on everything wasn't he! His zeta function, his sums!

There's only one way forward now--that's to let those benevolent CEOs at the AI companies democratize mathematics. They care about the intrinsic, and not extrinsic, value of mathematics. They care about the understanding! That's why they're rushing to publish these 250,000 page papers with their models, they're just so excited to share mathematics with the world.

That's the world I want to live in!

reply
Do you believe that researchers at historic, well-funded R&D labs (e.g. Bell Labs) should have kept discoveries private because they enjoyed resource advantages?

I am a researcher at an R1 university, and the nature of my work has changed, too, but the anxiety of working academics isn't a compelling reason to hide major results in physics nor math nor biology.

We ultimately do research to expand what is known, not to give ourselves satisfying puzzles or demonstrate prowess relative to peers

reply
> scoop Tristan Buckmaster

Probably important context: Tristan Buckmaster was collaborating with Levent Alpöge on Navier-Stokes before they were scooped by OpenAI. (Indeed, part of the controversy was that OpenAI wanted to include Buckmaster but exclude Alpöge from authorship).

reply
because they are not selfish and dont see the point of gatekeeping their work or credibility like so many other artists or professionals have been doing.

you can't win against LLMs best to find a field of work where you own things or whatever robots can't do like sales.

reply
I'm gonna assume robots are best equipped to sell stuff to robots.
reply
I was wondering how this would affect research incentives for difficult problems. If someone is about to embark on a long PhD process, but there is a chance that an AI swarm solves their problem 1-2 years in, isn’t that a bit risky?
reply
Yes, the whole academic system and structures and pipelines will have to be rethought from the ground up.
reply
Was it unfair in 1960 when very few people had access to a computer and almost everyone didn’t?

The first minicomputer DEC PDP-8 shipped in 1965. Apple II followed a decade later.

From today’s POV that’s a long time to wait, but I don’t see why AI shouldn’t be on a similar trajectory.

reply
Levent is a mathematician more than an AI scientist; probably he can not resist this, neither probably can say tsimerman at oai
reply
How about people who "align themselves" with CERN, LHC etc? So unfair!
reply
Terence is so far ahead of nearly all mathematicians, what big of a difference is it when an ai solves something?
reply
This is a good question. A big one is that Terence Tao only has 16 hours a day to do math, and he spends a lot of that time teaching math, giving lectures, and mentoring students. Tens of thousands of Claude agents can spawn in seconds and race ahead of any researcher.
reply
I don't understand your logic here. Won't these advancements in mathematical research advance the subject for humanity AND force newer mathematicians to leverage AI in their careers?
reply
This kind of proofs don't help humanity at all. AI uses a gobbledygook language that's incomprehensible by humans and the only outcome of its proof is a 0/1 result that you either believe blindly or not, for nobody can really verify it.
reply
This is not true. In the processes I've read about, AI produces natural language proofs first, which it then translates into Lean code which is verified by the Lean kernel.
reply
I read the Zulip chat and it sounds like sometimes the formalization is completed before the natural language proof.
reply
Another comment, I haven’t verified this, suggests this particular formalization is upwards of a few million lines of code. I really enjoyed this article from Wolfram where he argued that if we had a box that could prove anything it wouldn’t necessarily matter much because the important part is the application. Now AI might, will(?) provide both but when it comes to the proofs it’s hard to tell how useful they are _at this very moment_.

https://writings.stephenwolfram.com/2026/09/whats-the-future...

reply
AI only imitates natural language. Nobody can read and understand the 2500 pages of Claudese. Some may argue that AI can explain each part of the complex proof and therefore the entire proof is human readable, but if this was true, AI could produce a compact human readable proof in the first place.
reply
That assume a compact human readable proof exists in the first place. Can you prove that?
reply
I find it deeply discomforting that there are people like you who think mathematics is a personal race.
reply
I would like it not to be. My emphasis is on the mathematical community as a whole, not the individual mathematician who might have gotten there first if not for AI.

There's an alternate scenario in which the whole math community got to use AI to make progress on these problems, share their progress in Zulip chats, and learn from that process. Instead a few individuals within private companies and with vastly greater resources are doing it all themselves.

So I don't mourn who got credit and who didn't; I mourn the lost opportunities. The people who will walk away discouraged when they could have been brought along instead.

reply
deleted
reply
All academic mathematicians (and all(?) academics in general) are in a personal race, because their publishing record dictates their career.
reply
Today, yes. But academia cannot go on as it did so far. If it does, it's destined to irrelevance.
reply
Everything is a race to some extent.
reply
anthropic gains a lot by getting such news, billions
reply
[dead]
reply
At least, they could quietly collaborate with the community, and thereby wait until they have a more lucid, unambiguously correct proof, rather than dump a pile of slop on the community.

I can only imagine that the perceived financial pressures around the AI race/IPO caused them to announce this now.

reply
> There is an obvious cynical view, to do with the timing, but it is false, and I am pleased that I am in circumstances where it is obviously false.

FTA

reply
If only he could say why it's so obviously false.
reply
deleted
reply
deleted
reply
Collaborate how, exactly?

It’s basically impossible to do collaborative work with AI. This is a problem in anything ai does and it’s a huge problem we’ve had to deal with at my company. The only collaboration is between one human and the ai; a different model is required to enable more than that tbh.

And it seems like the author wanted to avoid dumping a pile of slop, AND explicitly referenced the timing of the IPO in the pdf here. Did you read it?

reply
That can be seen to be trivially not true.
reply
The polymath projects are a pretty amazing recent example of this. Also the recent AI boom in math has spawned many Zulip chats where people discuss these kinds of results and try to build off of them. There's one dedicated to digesting the Lean proof of Apoge's result.

> Did you read it?

Hacker News guidelines specifically call our this question as a no-no, but yes I did. I'm not calling into question how the result is being released. I'm asking if it should have been achieved at all before public availability of the model.

reply