I guess the real test of if this is useful is if Claude just effectively cracked RSA? Would be big news.
Maybe disproving the hypothesis would weaken some crypto but it has nothing to do with factoring primes. It’s about the distribution of primes.
No. I don’t think you understand what GRH is.
If anything the proof helps cryptography.
“I don’t think you understand what GRH is” and then proceeded to not explain what it is.
Why are you putting the burden to explain it on OP? He just pointed out the error in thinking this weakens encryption.

I mean, we’re still at 1 here, there’s a reason it’s outputting Lean code rather than actually doing any computation by itself (Because it can’t do that, it’s a text generator)
No this is just false. Easily provably false but I am not sure it matters here on Lemmy since it seems to be the goal to understand as little as possible then just talk out of your ass with authority.
I don’t think it’s true for everything on Lemmy, but people dislike AI so much on here that it’s true for AI. Even top tier models for now still screw up on some reasoning especially when it involves being reasonable and exercising judgment… for now… But when it comes to very concrete black and white stuff like math or finding programming vulnerabilities, AI is very good right now.
It can do multiplication like you and I can, it’s just not efficient,so why would anyone use it for that.
If you ask a local LLM to multiply numbers together without using any external tools, it will still get more right than the average person (who isn’t using any external tools, like pen and paper)
OK, each point in this list can be interpreted as “surely they won’t burn this much energy for another proof of an unsustainable path”.
Gold is edible. We don’t eat gold.
Gold is edible. We don’t eat gold.
Well, not a lot of it at least.
Oh you can eat a lot of gold it is just useless
Maybe I just enjoy the taste, sue me.
As it’s non reactive, pure gold is sadly tasteless
They still can’t ask as questions by themselves
Paywall 🤮
As a non-mathematics person, this still depresses me so much. It’s like like a kid finding out there is no santa. Or a religious person realizing there is no god. I can’t really quite describe what about this exactly depresses me, but I feel like ground has been taken beneath my feet. Not just this, the whole llm thing. Not because I’m afraid of losing my job or status. Just that the magic is gone makes me feel sad.
Its depressing for me because it means we will lose the knowledge and eventually just be operating completely out of the dark.
Let’s say 3 decades from now there aren’t any people employed in the field of mathematics, what will that mean? Well for starters it means that no one will have the skills to verify any hardcore mathematics outputs from AI. If AI says it solved something, everyone will just have to take it on faith. But it also means that there might not be any more research into mathematics at all in the not too distant future. Who is going to come up with the next problem to solve if there are no human mathematicians?
It’s very likely that in a post-AI world if we don’t all die from environmental collapse, we’ll live in a world of abundance with UBI. But people will probably still learn if they want to. AI is great at teaching. I just don’t see the problem. If all medical problems are solved, if people don’t age anymore, and lots of people want to just party, shop, and scroll Tik Tok, why should that be bad? And for people who are interested in science, they’ll be able to get more done so much faster. I also disagree with the idea that math will just be abandoned if there’s no financial incentive. Some people just like math. Some people just like science. Some people, even with UBI, would probably still do OnlyFans. People like different things.
The real asterisk in all of this is that we may actually end up dealing with environmental collapse. And it seems unlikely AI will be the cause. Rather, it’s been a long-time coming.
Why would AI waste resources on creating an utopia for humans if it could use those resources on itself?
Not to mention that resources are very much finite and if people stop aging they have more time to breed. If they have more time to breed, there will be more people which need to share those limited resources.
If it makes a difference, there is also a different, entirely human proof that was as I understand it, complete but not yet thoroughly checked/formatted/paper written, so had not been released yet. After Alpöge announced the AI assisted proof, it was hurriedly announced.
Unfortunately I rarely see people talk about it so it seems it didn’t get very much noticed
I agree. For me, I think it’s because the people who are pushing this by and large don’t give two shits about mathematics for its own sake. They don’t have any reverence for it, they are using this as a hype engine to convince more suckers to pour billions into their failing company.
They don’t care who they hurt along the way, the jobs or careers they destroy, what industries or fields of study they throw into turmoil. It’s the epitome of the, “move fast, break things” Silicon Valley, Capitalist Accelerationalist philosophy.
And on top of that, they got there by farming the totality of human knowledge, research, and open source software without and often against the consent of the authors/creators. Then they used that to create locked-down, proprietary, corporate LLMs and sell it back the the very people who it farmed that output from.
they aren’t just using it as a hype engine, they are using it because it’s one of the few textual tasks that is easy to confirm and is hard to do.
While yes, I was these days remembering how wonderful it was in the Web of my childhood, and then suddenly remembered that what was IRL was just as ugly as social media and all of tech industry now. It’s the Web that was nicer and far more intelligent.
The tech bros are to blame, but what they did was mainly one act - let in the baboons. What should have remained offline, came online. That’s all.
Now on LLMs - that was expected, honestly, after collecting enough resources and power attempting this was the logical next step.
It will bring havoc, but then it’ll get cold. Unless that kind of technologies is only available to a thin portion of the society. I fail to see how with any semblance of market mechanisms that can remain so (remain, because honestly I’m almost certain that before trying to sell that to us, governments and people of Epstein class and such did use them for like a decade).
Honestly, I think CS research has hinted we were going in this direction anyway. Dependently-typed languages like Idris and Agda have as term-inference for a while, it was based on exhaustive type-directed heuristically-prioritized search but it could still auto-complete/fill some non-trivial code, and Roq tactics could also make things that were disastrously wordy read much more simply and they are effectively a form of compile-time function (Elaborator Reflection is probably a better approach, but Roq’s two-language model is much more well-established).
Even without quite so interesting types, there were academic exercises like “Djinn” that could exhaustively enumerate lambda-calculus function bodies given the parametrically polymorphic type signature and a library of standard functions, and that’s been available for 14+ years? In an entirely different direction we’ve had SMT solvers can could be plugged in to a few languages to verify a LOT of properties that can be expressed in first-order logic.
Still, I think there’s plenty of magic, it’s just not in the places we can lay down (enough) axioms around symbolic structure / grammars.
Probably because human thought and intellect are the things that we, as a species, see as differentiating ourselves from any other animals on this planet. And so these intellectual and creative domains feel (and, in the past, were) exclusively the domain of humanity.
Automation is now taking that away.
What do we do when it seems like we, as a species, are on track to render ourselves useless?
I don’t know.
I wonder if this is what old world explorers felt like when the globe was fully mapped and there was no true frontiers anymore. What do you do when the thing that gave you a unique sense of purpose disappears in the face of technology and automation?
What do we as a society do when everyone experiences that at the same time?
A big difference between the explorers in your analogy and what’s happening now is that the exploration of the globe took generations to wind down, whereas AI has gone from chatbot to narrow genius in five years. (I say ‘narrow’ genius because the top experts in most domains still have a better intuitive understanding of their fields, and can still generally choose the path forward better than the machines.)
I keep thinking about the idea of future shock, which was popular in 1970s and ‘80s sci-fi;
a certain psychological state of individuals and entire societies, and a personal perception of “too much change in too short a period of time”.
I’m definitely feeling that now.
I, personally, would be totally fine to spend my days playing unproductive games. Just give me UBI.
Honestly, me too. But we both know that ain’t how it’s gonna go down…
“Damn it… I was hoping to solve that problem one day!”
Definitely not me, but some person, or perhaps a group of people. Like Wiles did with Fermat’s last theorem.
I guess I never thought human intellect was very special, and when I learned about neural networks, this sort of AI progress felt inevitable. It’s all still pretty mindblowing, but not completely inevitable
230 megabytes of proof. There is Santa out there for you. It’s no good, it’s some piece of slop that gives “ok” on a formal language interpreter. Even if there’s no bug in interpreter affecting that, confirming that those 230 megabytes have any relation to the task allegedly solved is a monumental human work.
This is nonsense.
Humans simply have to validate the question. The whole point of Lean and formal proof is to validate everything downstream of that.
The proof is real, and validated.
The monumental human work is reducing it and then understanding it. Although honestly LLMs will do much of this work too.
Isn’t this “Independent mathematicians who also used Lean software to check the proof that the author wrote/checked with Lean”?
It could still be valid of course, but I thought the proof was like 250000 pages and has only been looked at by the same automated framework. This is out of my wheelhouse and I don’t have a dog in the fight, but I’d still like some expert human eyes on it
They used 2 checkers that are maintained and developed independently of the Lean kernel. But they did use Lean formats/syntax. They’ve also done some manual verification, though there might be more to do. The proof did use the statement if RH directly from the well established Mathlib.
I’ve also seen claims that the proof has some weird things in it that might be hints that it is finding holes in the theory Lean models.
Honestly, even if it ends up not being a proof of RH, it will tell us something about computable / constructive maths.
It is 2028. Another lean CVE was registered; GRH was proven again. I am excited that yet another logic bug will be fixed.
Scott Optimized (Aaronson) declared the proof likely correct on friday, before this news, so I’d take their word for it.
Edit: Update from Scott Optimized:
[Update (Nov 21): The proof has now been verified independently by the Lean community, using con-leche, a piece of Lean that itself is formally verified all the way down to the axioms of set theory. This is a much higher degree of verification than a typical human paper accepted to a math journal. The proof is correct.]
i get what you’re saying but these automated proof-checking engines are solid. they don’t make mistakes. it’s like saying “ok so this compiler compiled 10 MB of C++ code into machine binary code, but i’d still like some human to confirm that it compiled correctly”. this is just not reasonable from a cost/benefit perspective.
what is interesing though is to look at the ideas nevertheless, but not to check whether every step was done correctly, but to figure out the important ideas and concepts used during the proof. maybe they can be re-used somewhere else?
There have been bugs in lean before. I think there still are known issues now. I recall specifically an “unsound type system”.
This reverification now though seems proven correct fully, according to Scott Aaronson
Truth is proof - different people have different burdens of proof. What use a proof that relies on faith in a system rather than mental construction?
They’re talking about “proof” in the math sense, not in a colloquial sense.
There are a bunch of ways that a proof written in Lean can be invalid.
The most obvious is that the problem statement is invalid or incomplete. This was not really a concern here as the proof used Mathlib’s built-in definitions rather than restating the problem, but nonetheless it has been verified.
Beyond that Lean proofs can contain custom axioms/assumptions, shortcuts, tricks, exploits. And there’s the possibility that the Lean compiler used in the proof had a bug.
Anyway all that has been tested and verified. The problem statement is valid. The proof doesn’t contain any tricks or assumptions. It compiles on every major Lean kernel.
Basically: it is almost certainly mathematically true. The next step is humans actually understanding it.
The next step is humans actually understanding it.
So what we got here is “42.”
“Claude, explain this proof to me in a way that I can perfectly comprehend in totality without requiring me to understand any mathematics beyond high school. Make no mistakes.”
I’d settle for a human mathematician understanding it. What good is a proof that only other software understands? Am I missing something? Does it give us some sort of useful or more widely applicable formula or constant?
I can see LLMs doing a decent job at that.
It’s 250,000 pages of explanation for 17 million (I think) lines of lean code.
It’s incomprehensible, and therefore probably of little value.
It’s incomprehensible as a whole but not in pieces.
It will be simplified, refactored, maybe rebuilt from the ground up. It won’t be incomprehensible for long.
The last I looked the size had already been reduced by 25% based on community efforts.
Kind of fun watching the Lean Prover Zulip chat play with it.
Is it even possible to distill this down to the type of proof a human would produce that is orders of magnitude more concise and has coherent meaning, or is it just largely slop that would be easier to start from scratch with the monumental task of a human-made proof than it would be to painstakingly unravel it?
If scientists are actually able to independently verify the proof, then it must be understandable to them.
I wonder how many scientists it took to independently verify 230 megabytes of proof. And in a week no less.
And if it’s not a minimized proof that weighs 230 megabytes, then whether anyone has been able to extract that.
Probably some of both. Though, given the short history on these proofs, it will likely be more “slop” (or theft) than anything else.
While this is orders of magnitude larger, the first 4-color theorem theorem had a lot of cases that were quick for a computer to exhaustively check. But, it turns out a lot of those cases are better explained without exhaustive checking, so the currently taught / best proof of the 4CT is shorter and has less long, mind-numbing checks.
It’s entirely possible that there’s actually no more “insight” from the LLMs, and they were just better able to generate the massive type signatures and terms that Lean needs to begin mechanically checking, because they are able to generate and discard good syntax much faster. But, there might also be some interesting bits that people (or LLMs) will be able to extract/abstract/generalize/shrink into something humans will find insightful. I think the Navier-Stokes related proof did have some stuff in it that humans found interesting beyond the final result.
Bless you
Bless you for blessing him 🙏
On Friday Anthropic mathematician Levent Alpöge claimed to have used the company’s Claude AI model to solve the Riemann problem, perhaps the most important open quandary in mathematics.
“We have full confidence that Alpöge’s proof of the Riemann Hypothesis is correct, and we congratulate him on this achievement,” mathematicians wrote in a joint statement by the Lean and Mathlib communities
Are we calling it “Alpöge’s proof” despite the fact that he didn’t come up with it?
if the human drives you from A to B, do we give credit to the driver or to the car?
I think it’s unclear how the LLM was involved, I doubt the LLM came up with it itself. It’s a tool and assisted him but he ultimately is responsible for it. If it were garbage he’s the one who will take the reputations hit.
I doubt the LLM came up with it itself
In math, these things are attributed to those who can actually complete the proof. Maybe this could be called Alpöge’s hypothesis of the Riemann problem’s proof or something if he did come up with the idea.
I don’t think ownership is determined by whose reputation would take a hit. I mean, at this point, the Lean and Mathlib communities would also take a reputation hit if it was proved wrong, but it doesn’t make sense to name it after them.
Yep, Alpöge should take the responsibility, and should not be allowed to have authorship on the result (as previously ruled so).
Maybe we should just call any proof that uses calculus Newton’s proof.
He has a writeup describing some of the process: https://alpo.ge/remark.pdf
You think Claude should be credited instead?
Maybe all mathematicians that ever published anything should be credited? After all, the plagiarism machine can only predict characters based on what it’s been trained on.
Claude isn’t a person, so no. I think if Claude did the important work, then it shouldn’t be named after anybody.
The proof just… showed up, then? Nobody made it?
You don’t HAVE to name mathematical objects after a person. We could just call it Riemann’s Theorem (a proved hypothesis is a theorem) and include the date and by-line if we need to reference a specific one in (e.g.) a bibliography.
Decolonize maths!
Nobody made it?
Is that so weird, though? I’m drinking soda from a can that nobody made. My desk is covered with things that nobody made. Bots make tons of stuff and we don’t name those things after the bots or their operators.
There’s no reason to think that a “body” has to make something just because it exists and is complicated. That’s the same mistake that creationists make instead of understanding evolution.
But that’s all off-topic. The question isn’t where it came from, but who it is named after.
That’s already long established. It’s the Riemann hypothesis, it’s right in the title.
I’m not sure what you’re suggesting at this point. When it’s published in a journal are you suggesting the byline should be left entirely blank?
You have completely forgotten the context for these comments, so rather than asking me to explain things, I think it would be more straightforward for you to just go back and read the comment chain.
Alright, so should it have been:
“We have full confidence that nobody’s proof of the Riemann Hypothesis is correct, and we congratulate nobody on this achievement,” mathematicians wrote in a joint statement by the Lean and Mathlib communities
Because that seems quite unhelpful to me.
Can someone ELI5 the implication of this proof?
We’ve confirmed something that was long suspected to be true, but learned nothing in the process. Hard to say if humans can even understand the proof. It’s true… and that’s it.
Usually the importance of a problem like this is that the journey to solve it leads to major breakthroughs that can touch other fields of mathematics. So it’s a bit surreal to see the most famous unsolved problem finally felled, with the caveat that we learned nothing in the process.
In the process we also confirmed something that was long suspected by many not to be true.
That AI could solve the RH within a few years. Tao just months ago was saying he only thought AI would be able to solve it but around 2050.
We’ve come a very long way from “LLMs won’t generalize” which was the predominant belief a few years ago.
So the proof reflects a lot more than just the rate of primes.
Ha ! So this is a good starting point for a science fiction story where an artificialy intelligent system creating this proof learns something that gives it an edge over the human specie.
The short answer is “not much.”
Us assuming that the Riemann hypothesis is true is the basis of much of our mathematics. If it was proven false, it could affect a ton of the math we’ve built on top of it for centuries, but it being proven true is just confirmation of what everyone was sure of but couldn’t quite prove.
math we’ve built on top of it for centuries
Well, almost centuries at least. The Riemann hypothesis is less than 200 years old (167)
It’s counter proof would have been monumental. This is big. Perhaps the biggest of ours and a few generations surrounding, but not practically so.
It changes little. It’s the result we’ve come to expect. Some other conjectures will use it, maybe Goldbach.
But the disproof would have hurt a lot of number theory, maybe engineering, and probably crypto.
Could primes being predictable result in a weakening of the design of encryption because it would be easier to calculate large primes?
No, not directly at least. Encryption relies on the difficult nature of factorisation, not knowing if a specific number can’t be factorised (is prime).
- Possibly some of the internal components of this could be extended to help factorisation problems (but also possibly not)
- Most modern encryption algorithms moved away from primes because of other weaknesses like quantum factorisation anyway
So there are prime numbers, that are only divisible by 1 and themselves. Here’s a set of all the prime numbers bellow 100:
[2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89, 97]
As you start getting larger and larger numbers, primes show up less and less. The Riemann hypothesis is that there is some pattern, some formula to the distribution of prime numbers.
So if there is a pattern and you had the formula, you could predict which arbitrarily large number is going to be prime. Otherwise, you just have to find a big number and check if its divisible.
My understanding is that it is not a pattern or formula that helps finding primes. Just a more accurate estimate of how many primes exist that are smaller than an arbitrary large number. Still need to do the same work as before to check your large number to see if it is a prime.
The other poster said a ton of math is built upon that idea? Like what and why?
Sorry I know asking this is alot, im just curious.
The Wikipedia page lists various consequences. Mostly things that will not mean anything at all to a layman. The other thing is that claims that “all of modern mathematics” hinges on the Riemann hypothesis are extremely overblown; it’s mostly consequential to number theory, which is something many modern mathematicians work on/intersect, but there’s also many who never touch it and for whom the Riemann hypothesis being true or false changes nothing.
We love using big primes for public key cryptography.
That’s some black magic math that lets you take a pair of big primes. Keep one for yourself, give the other to the world. You can then encrypt or sign things with your number, and everyone with the public prime can decrypt or verify that you signed it. And vice versa, people can encrypt things that only you can decode.
As I recall, it all relies on the fact that it’s a lot of work trying to find suitable primes for this. Knowing the pattern to how often primes occur can make it easier to break. This proof only predicts how many primes we’ll find, so it’s the basis for it.
Big primes are made with that Fermat theorem though, it has nothing to do with their distribution.
I wish I knew for sure, but I don’t.
Best guess, I can say that a lot of mathematical profs work by looking for the simplest description, then building up from there. Primes are as mathematically simple as it gets, so its kinda important to keep in mind there’s “simple” numbers hiding out there in the unthinkably large digits.
Does this mean I can conjure up a crazy prime number with a little algebra now?
Riemann isn’t a prime generator, it’s a prediction of how many primes exist in any contiguous sequence of integers.
What if you binary search on that sequence? How much computation is required to predict the number of primes?
Largely unrelated. Generating primes has always been relatively easy. All you need is a random number generator, a primality checker and the ability to keep adding one until the primality checker says it’s prime.
Now if you want to count the number of primes below a certain number, you’re moving closer to Riemann Hypothesis territory, but I don’t think this result will help much with that either.
No more so than you could do before - we’ve been assuming that Reimann is true since forever, a lot of modern mathematics depends on it. And we already have ways to conjure up extremely large numbers that are almost certainly prime, such as Fermat’s primality test. RSA encryption depends on us being able to do that, in fact.

?
Reman hypothesis
The nerds have had their run. Time to give the jocks another turn and see how things go.
People who don’t need to put their dick everywhere because they already did… in high school.
Or maybe the artists. Let’s have them run things.
Are you like 8 years old?
No
I really do not know what kind of person would comment such an ugly piece of text, but you should realize that ALL of this is the work of nerds.
And not all of them have dicks.
And some might not want to fuck everything that moves.
And most of them have a completely normal social life the moment you remove the jocks.
It’s an extremely bizarre comment
If someone figures out how to set up a fediverse account, I hate to break the news, but this is 110% nerd activity. We’re all nerds here.
What the fuck qualifies as a jock these days anyway? Does anyone still use that term unironically outside of a Hollywood writer’s office?
You’re good at sportball? You learned all those sportball rules? You keep a catalogue of all the best sportball players? Good job, nerd.
I seriously can’t even tell what point they’re trying to make































