674 points iamronaldo 1 hour ago 376 comments

jrflo 1 hour ago | parent

Insane if true! Would also be great for Alpoge after being scooped by OpenAI on Navier Stokes...

Edit: just saw the Lean code is included in the footnote. Considering the author's reputation, I think it's fair to remove the qualifications from the post title, by the standards of today, it's proven.

hgoel 50 minutes ago | parent

This may be how the court of public opinion often works, but plenty of otherwise very prolific mathematicians have claimed to solve problems only to turn out to have been mistaken.

jrflo 48 minutes ago | parent

This is true, but since there is a lean formalization the odds of that are much lower. Only the correctness of the problem statement needs to be verified, which I would assume is fairly easy for RH? But I am not an expert, so would be curious to hear from someone who is.

hgoel 35 minutes ago | parent

I'm not an expert in this area either, so I too will be waiting for experts to review things, but I will remain skeptical until then.

After the excitement of the Navier-Stokes solution, only for it to eventually seem to not really solve the full problem and leaving discourse about the problem in a very muddy state, I'd rather wait and see if it turns out this proof also makes some assumptions that ultimately limit its impact.

It is probably still some form of breakthrough, but I will expect it to not be as big as a full general proof of the Riemann Hypothesis, until proven otherwise.

Edit: A thought though, if this proof is correct and can't be made much shorter than the current ~250k pages, I wonder if the result has more value as a sign that human accessible proofs to these kinds of hypotheses were stuck behind some mathematical abstraction/tool that we haven't quite discovered yet? Sort of like how you can brute force algebra with real numbers for working with AC circuits, but it's pretty ugly and painful compared to using phasors. If so, maybe the real contribution from AI-driven high profile proofs would come from mathematicians studying the proofs and devising tools to make them more accessible.

azan_ 29 minutes ago | parent

> After the excitement of the Navier-Stokes solution, only for it to eventually seem to not really solve the full problem

The version of solved problem is exactly the one that's permitted by Clay Institute. If it wasn't done by AI there would be no such discussion and everyone would just accept it as a huge breakthrough.

yewenjie 1 hour ago | parent

> The argument is incredibly long at the moment, and while its formalization is finished and it is routine to produce a huge, unreadable document containing sufficient detail, in my view of my path I have just taken a step. I intend to do my best to understand and impart understanding of this argument and its certain simplifications, to speak about it, to write it properly.

Okay, I am anxious whether this holds true or not, but he claims that the formalization is finished.

Why not just release the formalization in the meantime?

This is probably the biggest day in math in recent memory.

Edit: nevermind, the Lean formalization is linked.

csheehan10 1 hour ago | parent

He has linked to the Lean formalization in this note.

jamessb 1 hour ago | parent

> Why not just release the formalization in the meantime?

I think this is what is linked to in footnote 1, inside the 14-part tar.gz archive.

asa123 57 minutes ago | parent

doesnt he link to a lean file in the pdf?

scottmsul 54 minutes ago | parent

He did, see the links below.

https://alpo.ge/lean.tar.gz.part01 ... https://alpo.ge/lean.tar.gz.part14

Though he decided against releasing a slop-filled pdf, which would have been over 250,000 pages.

programjames 53 minutes ago | parent

Biggest day so far.

mellosouls 40 minutes ago | parent

<pedantic>

Implied in "recent memory", which by definition includes everything up to the point of that comment, but not beyond.

</pedantic>

krona 52 minutes ago | parent

> Why not just release the formalization in the meantime?

Nobody has a machine with 10TB of RAM required to run the Lean proof?

josefx 30 minutes ago | parent

Throw some swap space at it?

jmyeet 49 minutes ago | parent

I'm old enough to remember when Andrew Wiles announced a proof of Fermat's Last Theorem way back in 1993 [1]. It was long and complex and took at least 2 years for the maths community to accept it.

I wouldn't be surprised if this isn't similar (or worse). The days of short, succinct, elegant, intuitive proofs is probably over.

[1]: https://en.wikipedia.org/wiki/Wiles%27s_proof_of_Fermat%27s_...

ilaksh 6 minutes ago | parent

On wikipedia it says it's 250,000 pages. That might be an error on the page though.

xianshou 1 hour ago | parent

This is Levent Alpöge (https://en.wikipedia.org/wiki/Levent_Alp%C3%B6ge), a supremely qualified mathematician and one of the best in the world at guiding AI towards new discoveries in mathematics.

If he says he has Riemann, well then, he has Riemann.

charles_f 58 minutes ago | parent

Q.e.d.

solyin 56 minutes ago | parent

Quod Erat Demonstrandum, meaning "that which was to be demonstrated" or "what was to be shown"

away0g 51 minutes ago | parent

TBD

bitlax 45 minutes ago | parent

Wow I just thought those were the initials of the guy who wrote every proof ever.

mockerell 41 minutes ago | parent

Don’t forget his French cousin CQFD

ben_w 31 minutes ago | parent

C'est quatre fromages dauphinoise.

bee_rider 33 minutes ago | parent

This would be a good Easter Egg for a story. Mathemystics who worship the ancients, Pythagoras and the mysterious QED.

mikrl 18 minutes ago | parent

Characteristic of the disciples of QED was their renunciation of proper names in favour of a black rectangle.

igleria 38 minutes ago | parent

We used to joke at university that it meant "queda entonces demostrado"

darkwater 32 minutes ago | parent

I believe it's actually the correct "reverse acronym" in Spanish? I mean, keeping the Latin one they created a Spanish version that brings the same exact meaning, in Spanish.

In Italian we use "CVD" for it (Come Volevasi Dimostrare)

eriquesito 32 minutes ago | parent

Isn't it what it literally means in spanish?

layer8 36 minutes ago | parent

You don't say.

ilaksh 13 minutes ago | parent

He did post the Lean proof.

Also the Wikipedia says that the full written proof is 250,000 pages. Which I am guessing might be a mistake. If not it could take a while for humans to verify it.

well_ackshually 9 minutes ago | parent

Anthropic (and their researchers, by extension) have an habit of taking a massive shit on the entire mathemetical (or physics, or their current subject) community, along with disdain in their messages that consists in 50% of "trust us bro" and 50% of "verify it yourself lol we ain't doing this shit". All the while reaping weeks of dick sucking. Even if their discovery happens to be wrong. (lol navier stokes)

Mathematicians aren't protected from AI psychosis. Mathematicians that think that 250k pages is anywhere close to being mathematics are drowning in the psychosis, and making the fields worse as a whole.

bonzini 7 minutes ago | parent

He says it in a footnote, probably referring to ~2M lines of Lean?

simonw 56 minutes ago | parent

Levent was one of the two mathematicians working on Navier–Stokes when OpenAI published their work on that problem which was produced in a rush when they heard a rumor that someone from Anthropic had promising results on a millenium prize problem.

QuesnayJr 21 minutes ago | parent

It's a sign of the times that when someone with an AI beats you to one Millenium Problem, you can just turn the attention of your AI to another Millenium Problem.

finghin 11 minutes ago | parent

It's almost certain that they have had people working on Riemann since way before the Navier-Stokes theorem proof was nearing completion. It's a long shot, but the value of a proof is next-level compared to NS

teekert 12 minutes ago | parent

I think he worked together with this person? https://www.youtube.com/watch?v=Mzbtj5nkMXI

Pretty crazy interview, he didn't want throw the Anthropic employee under the bus when OpenAI offered co-authorship on the Navier Stokes paper. He suspects OpenAI somehow used his codex usage btw. Almost unprovable of course.

simonw 9 minutes ago | parent

Yes, it was a joint effort with Tristan. Part of the scandal around it came when OpenAI offered to let Tristan help write up their result but wouldn't allow Levent to be involved because he worked for their competitor.

keybored 56 minutes ago | parent

* Levent Alpöge

harry8 52 minutes ago | parent

Whose loss did he steal to get there?

Sorry, Anthropic have earned their reputation on this all by themselves.

epistasis 48 minutes ago | parent

Anthropic? What incident(s) are you talking about? I'm only familiar with OpenAI doing this to a duo, then offering credit only to the non-Anthropic employee.

stymaar 39 minutes ago | parent

the Anthropic employee from the duo in question was the same Levent Alpöge featured here.

p-e-w 31 minutes ago | parent

Yeah, I’m sure all that stuff was just waiting in some grad student’s notebook. 2.5 Millennium problems solved, ETH disproven, hundreds of other mind-boggling results, but obviously this all just happened because people uploaded their secret breakthroughs to ChatGPT.

As a child, I used to laugh at the story of Columbus and his egg, now I get to see it re-enacted in reality! “We could have done the same thing…”

jascha_eng 46 minutes ago | parent

I mean a single person can always make mistakes regardless of how smart they are. But this does make it sound a lot more exciting than if a random vibe-prover came up with it.

jrootabega 33 minutes ago | parent

No theycan;'t

niksmather 28 minutes ago | parent

Brother it might be worth looking up Andrew Wiles

igleria 39 minutes ago | parent

Congratulations on making my palms sweaty. Fields medal for Levent/Claude...

talon8635 32 minutes ago | parent

You mean, if the AI says it does, than it does. And it’s probably right, because it’s smarter than all of us and we basically just have to trust it now, which is exciting and frighting in many ways.

fearmerchant 28 minutes ago | parent

Thanks, the title made it sound like the custodian walked by and typed "solve this" and found an answer.

stavros 20 minutes ago | parent

Man, Good Will Hunting 2 is much more boring than the original.

rfgplk 13 minutes ago | parent

That's pretty much what he did regardless. I can assert with absolute confidence that he read effectively 0% of the "proof".

hersko 11 minutes ago | parent

How could you possibly know that?

_fizz_buzz_ 7 minutes ago | parent

Huh, what a conincidence that the AI presented the proof to Levent Alpöge (a top class mathematician) and not to some Rando McRandomface then? What are the odds?

xyzsparetimexyz 21 minutes ago | parent

What do I have to do to get an asteroid named after me?

amenghra 4 minutes ago | parent

Change your name to 354 Eleonora ¯\_(ツ)_/¯

heisgone 20 minutes ago | parent

Let's redefine AGI as solving every known problems to man so we can stop moving the goalposts.

yesbut 17 minutes ago | parent

it hasn't solved any of mankind's problems yet.

DFHippie 14 minutes ago | parent

It will need cooperation from mankind. So do we.

aeonik 12 minutes ago | parent

It's solving tons of my problems every day...

win311fwg 7 minutes ago | parent

Hence why the previous comment expressed "mankind", not "HN accounts".

lionkor 3 minutes ago | parent

You can have the AGI label, you can put it next to the "actually, we've solved world hunger, people just aren't working hard enough" discovery, somewhere on the shelf with "every billion I make trickles down to the barista when I get a coffee at starbucks".

The AGI label doesn't matter. You can call it AGI when you feel like it's generally intelligent--I've yet to see any model that even remotely seems intelligent, but it all depends on where your bar is.

croes 20 minutes ago | parent

> If he says he has Riemann, well then, he has Riemann.

That’s not how this works, especially if he works for Anthropic

artninja1988 18 minutes ago | parent

Is it really wise to only rely on credentiallism to verify things? Plenty of mathmaticians top of their fields believed they have proven something when the proof continued some flaw. Just look at Mochizuki

317070 16 minutes ago | parent

credentialism, and a lean codebase that compiles and certifies Riemann is correct.

well_ackshually 12 minutes ago | parent

No. A lean codebase that certifies that whatever definition of the Riemann hypothesis their LLM wrote is correct.

I can write Lean that certifies that 2 + 2 = 5 is correct, by redefining every axiom about natural numbers on the way. Or abusing bugs in Lean. Stop believing Lean is the response to everything.

questionable_ju 14 minutes ago | parent

he's just pointing out that this isn't some rando claiming to have video of aliens.

lanstin 11 minutes ago | parent

Math is hard; even to understand the questions and what the entities are which the questions address takes a lot of deep study, even for smart people. So credentials are not sufficient but they are for sure necessary. I have tried my whole life as a software engineer to study math as an amateur and then started a masters program and learned more details thoroughly in three months than in decades of self teaching.

dgellow 5 minutes ago | parent

No it’s not. HN main approach to analyzing anything is to rely on chaining heuristics. Which is generally sort of fine for software, but doesn’t make sense in the real world. It’s at best one signal to weight in your analysis, nothing else

some_furry 17 minutes ago | parent

> If he says he has Riemann, well then, he has Riemann.

Yeah, no. This kind of framing is inappropriate. Let the mathematicians check the proof.

Bluestein 12 minutes ago | parent

  You are absolutely right to push back!

rfgplk 14 minutes ago | parent

I don't understand this comment. For instance,

> a supremely qualified mathematician

How? No matter how "qualified" you are, the AI is still doing ~100% of the work. And since we're talking about superintelligence (Levent's intelligence hasn't even been stated anywhere), it's questionable whether a human can even "verify" or "check" the AI's work.

_fizz_buzz_ 10 minutes ago | parent

- It's a lot more straightforward (easier) to verify a proof than to come up with a proof

- It is also unclear if this proof is something a human couldn't have come up with

taegee 11 minutes ago | parent

> and one of the best in the world at guiding AI towards new discoveries in mathematics.

You mean repeating "keep going" and "believe in yourself" over and over again?

https://www.anthropic.com/research/riemann-zeta#claudes-meth...

Philpax 10 minutes ago | parent

Believe it or not, "Jarred Sumner" is not the same name as "Levent Alpöge".

duskdozer 7 minutes ago | parent

wow.

auntienomen 7 minutes ago | parent

Alpoge is a smart and principled guy, and he is probably the world leader at getting LLMs to do math. But he's not an expert in these subjects. If we're going to argue from authority, let's wait for an actual authority to weigh in

hmokiguess 7 minutes ago | parent

If only we stopped with the marketing and named the person, so sad to see someone like him reduced to "Anthropic Employee"

racl101 5 minutes ago | parent

Wonder what his intellectual peers think.

gus_massa 1 hour ago | parent

It's not a random Anthropic employee. It's claimed by Levent Alpöge. The guy is on fire and already has like 3 or 4 very interesting verified result this year. I've showing for months his wikipedia page to my coworkers in the math department in my university.

10111001 56 minutes ago | parent

> The guy is on fire and already has like 3 or 4 very interesting verified result this year.

Actually Claude is on fire, and this guy is along for the ride:

> Claude and I have had a great time together with such breadth in recent months, but this time Claude far outpaced me near the finish line. No, Claude flew, and danced, and sparkled near that finish line. I had us on the line of combining two quite different ideas present in a toy setting, one of which was already new and shocking, making very simple one of the shadows of Riemann over which we mathematicians felt most triumph. I guided us to the right situation analogous to that toy setting, and then Claude began to leave me in awe.

The anthropomorphization is cringey.

plutokras 52 minutes ago | parent

Someone has to guide Claude still. Otherwise we would see layman bragging on LinkedIn about cracking random math problems.

gus_massa 23 minutes ago | parent

I've seen post here of people solving their pet math problem. They are easier than the RH, but they had that as a personal project for years and now they solved it with Claude or ChatGPT.

For example, https://news.ycombinator.com/item?id=49961538 that uses the RH to prove that the sum of some subset of the primes is finite. [Congratulations jdb1729, now you can update the solution to an hypothesis free solution.]

dexterdog 51 minutes ago | parent

This guy knows his stuff. Along for the ride would be if I made this claim.

The second part sounds like the timing of this was to pump the IPO.

artninja1988 30 minutes ago | parent

This has to be the most impressive result in order to pump an IPO in human history, right?

meowface 47 minutes ago | parent

"The anthropomorphization is cringey."

No it's not. If Claude just solved the Riemann Hypothesis (if that is later proven to be the case), the only cringeworthy thing is your response to his completely understandable description of the events.

albatross79 8 minutes ago | parent

Yes, it is.

azan_ 46 minutes ago | parent

Why is it cringey? Sounds like a good and natural description to me.

nozzlegear 19 minutes ago | parent

They told you what's cringey: the anthropomorphization. "Claude and I had a great time" – no, Claude didn't have a great time, it's software. Levent Alpöge had a great time using Claude.

gus_massa 45 minutes ago | parent

My guess is that the guy has a ×10 discount to use Claude [1] and also he is ×10 better at Claude whispering, so he get like ×100 better results than us mortals. [2]

[1] So he spend the equivalent of a lot of money, but not milllions.

[2] Mathematician here. I worked with people that is much better than me with pen and paper. I guess he is much better than me steering Claude.

PS: When we talk about Claude, we call it "Claudito" that save us the effort to pronounce it correctly in the middle of a sentence in Spanish.

trio8453 45 minutes ago | parent

Cringe is in the eye of the cringer.

jasonlotito 43 minutes ago | parent

> The anthropomorphization is cringey.

https://www.usni.org/magazines/naval-history-magazine/1998/d...

> Actually Claude is on fire

The anthropomorphization is cringey.

daemonologist 38 minutes ago | parent

The writing is certainly exuberant, and we are used to cringing at this kind of language because we are used to it being unjustified. But if the proof holds up then I can hardly object.

Sopel 27 minutes ago | parent

I think he's just happy to finally be able to talk to someone(thing) more intelligent than himself.

origini 24 minutes ago | parent

Why srsly?

If Claude is breakthrough for Riemann, can't we give people slack on anthropomorphization?

People do this with animals, people do this with objects but an AI bot, writing like a human this is cringey?

albatross79 9 minutes ago | parent

Using "srsly" is also cringy.

teekert 52 minutes ago | parent

I guess this is what you get when you pair top human intelligence with top machine intelligence (as in access to insane compute + latest models).

xrisk 45 minutes ago | parent

we’re in the human + chess engine can comfortably beat chess engine era.

lordnacho 40 minutes ago | parent

Which is interesting, because it's not necessarily the best chess players who are the best computer-assisted chess players, IIRC

xrisk 36 minutes ago | parent

the best chess players today are engines, not humans. that’s the competition. the best computer-assisted chess player is probably another computer.

kypro 32 minutes ago | parent

And none of the best computer-assisted chess players are better than AI today. In the same way a human + chimp pilot is not going to beat a human pilot.

This era of human relevance in fields like maths will end – probably in a month or two if we're lucky.

lordnacho 29 minutes ago | parent

I suppose we are already in the final stages of human-assisted mathematics, is what you're saying?

charlieyu1 19 minutes ago | parent

No, humans can always invent new fields and ask new questions

xrisk 21 minutes ago | parent

that’s not how math works.

teekert 8 minutes ago | parent

Exactly. We don't even understand all of the new solutions yet, I heard Terence Tao say he had only looked at about 10 of the OpenAI proofs, found some good things, bad things, mistakes. But it will be his (and other's) understanding of the proofs that will give new insights.

In other words, what will the AI solve if we don't even have the language yet to formulate its next challenge?

The AI would need motivation... (Please no!)

mb7733 23 minutes ago | parent

What is a computer assisted chess player today? Why would they do anything besides play the engine's move?

triceratops 12 minutes ago | parent

They don't use the engine while playing. They use it in training, to learn and try new moves.

jkhdigital 44 minutes ago | parent

This. Frontier AI means that the sharpest minds in the world now have an infinitely patient and motivated colleague to work through every idea.

I’m not claiming to be “top human intelligence” but I’ve published in premier venues in my field and I can say confidently that none of my human colleagues can keep up in conversations on my research niche like Claude can.

puff_pastry 1 hour ago | parent

Woah, if this proves to be correct it'd be massive, for maths, computer science...and AI too, I guess!

The worry I have right now is this though:

> The argument is incredibly long at the moment, and while its formalization is finished and it is routine to produce a huge, unreadable document containing sufficient detail, in my view of my path I have just taken a step.

this feels like we're moving away from the old world of 'science', where we could actually read an experiment, understand its steps, and, with enough time, possibly reproduce it.

And we're back in the world of 'magic' and 'belief'. Yes, there are formal ways to prove that this is correct, but if the proof is so convoluted and hard to read, is it actually something a human will ever find the time to go through and understand? We're going to start delegating (oh well, we're already doing that, I suppose) the understanding to AI, so it can just help us condense it down to smaller pieces that are digestible.

Still, very strange dizzy feeling, much more than Navier-Stokes

Waterluvian 54 minutes ago | parent

My perception of these maths solutions is that we're decoupling two related but distinct elements: the answer and the understanding. We have a big machine that tells us the answer is "42." That's useful. We know the question is answerable and the general shape the answer takes.

I'm just not convinced that we're swapping one world for the other. We're getting a hint that will probably help us enormously with what was always the main objective: understanding the result.

valsann 30 minutes ago | parent

well, "what is 6 times 7?" and "how many roads must a man walk down?" are wholly different questions ;)

Waterluvian 23 minutes ago | parent

Are they, though? Homer (the philosopher not the poet) already answered the latter for us: https://www.youtube.com/watch?v=PY3fqF6-jyE

jacquesm 5 minutes ago | parent

Six times nine, actually.

magimas 54 minutes ago | parent

> this feels like we're moving away from the old world of 'science', where we could actually read an experiment, understand its steps, and, with enough time, possibly reproduce it.

> And we're back in the world of 'magic' and 'belief'. [...]

I don't know, to me it feels more like this might be the job of future mathematicians?

AI proofs noone understands are good, but maybe don't lead to that much. So maybe in the future, mathematicians will be more like "data miners" or more like experimental scientists, where they take a statement already known to be true (in the case of experimental scientists that "statement" you know to be true would just be nature itself) and try to find ways to make it accessible to human knowledge.

ballmerpoint 36 minutes ago | parent

> I don't know, to me it feels more like this might be the job of future mathematicians?

Nobody would pay us mathematicians enough to prove it in the first place; why would they pay us enough to explicate it? I doubt there’s much value (in the sense of capitalism) in an explanation. Society by and large does not value pedagogy, or education, or even writing.

The social structure that trained a small number of research-level mathematicians every year is evaporating, and I doubt anyone can stop that now.

kjanus03 23 minutes ago | parent

> Nobody would pay us mathematicians enough to prove it in the first place;

weird thing to say about a problem which literally had a $1 million dollar prize attached to it but fine

jubilanti 14 minutes ago | parent

Exception that proves the rule. Try making a living working off these big prizes. It's the same with infosec and bug bounties on a different scale. You have to already have a secure income, as the prize is a bonus at the end. In what other industry would you work for years on something that inherently can never return a profit, other than the chance of a prize at the end for the one person who got their first.

Society doesn't pay for math labor, it pays to celebrate the top 1% for labor subsidized by other institutions.

ballmerpoint 12 minutes ago | parent

You missed the word ‘enough’. One million was obviously “not enough” (nor was that specific one million ever paid to anyone — it represents an investment of zero in that sense).

It took on the order of ten billion dollars to develop Claude, depending on how one does the accounting.

IsTom 8 minutes ago | parent

I'd wager that Anthropic has spent more than $1 million on solving this.

Rzor 32 minutes ago | parent

It's all tech priests all the way down then. Either that or the Bicamerals from Peter Watts' Firefall are about to rise. Maybe both.

xyzsparetimexyz 51 minutes ago | parent

Can't we untangle these proofs? How analogous is it to 'refactoring' the proof to reduce the amount of duplicated code etc?

StilesCrisis 18 minutes ago | parent

You could, just like you could slowly disassemble your house back into individual pieces of lumber and nails and drywall, but it would be laborious and at the end of it you'd just have a pile of semi-broken construction materials.

xyzsparetimexyz 16 minutes ago | parent

I believe that it could be more akin to decompilation? When done correctly, you turn something a bit crude into potentially beautiful.

lordnacho 50 minutes ago | parent

True, but that's how most of humanity feels about any discovery. You can't come close to understanding it yourself, you are just trusting there's a process that produces someone who says he understands it, and there's hundreds of years of institutional harness for us to believe that he has actually understood it.

Now that someone is an AI.

keybored 49 minutes ago | parent

AI is inherently about getting cooked/digested answers. It’s already finished. Nothing to build on.

But I still wouldn’t call the end result Magic as long as you have something to hold on to. I can see that this building in front of me (waves hands) is upright. Okay, now I remove this beam. It collapses. I don’t care about the whole structure. Just that it stays upright.

Giving up understanding is not the same as blind trust. Or maybe I’m too philosophically naive in this area...

BearOso 42 minutes ago | parent

> And we're back in the world of 'magic' and 'belief'. Yes, there are formal ways to prove that this is correct, but if the proof is so convoluted and hard to read, is it actually something a human will ever find the time to go through and understand? We're going to start delegating (oh well, we're already doing that, I suppose) the understanding to AI, so it can just help us condense it down to smaller pieces that are digestible.

If a tree apparently falls in a forest and no human goes to verify it, did it really fall?

No matter the verification results, as with most theories, for practical uses it was already assumed the hypothesis held. For mathematicians, they'll just say "Oh, it is true, but what's this proof? It's not understandable, so there's still a miniscule chance it's wrong, and we can't learn anything from it, so what's the point?"

ralusek 39 minutes ago | parent

What if this is metaphorically like a P = NP scenario? Working out the solution to these problems might be the "NP" here, but it remains sufficiently tractable to understand the solution in "P." So human understanding could come along for the ride, at least for a while longer.

bitwize 38 minutes ago | parent

I'm of the belief that "Claudese" with its load-bearing seams is just the models evolving our language beyond our capacity to use or understand it. Read up on your Continental philosophy to understand why and how: there's no such thing as a fixed "meaning"; meaning is something constructed for the immediate need and then discarded or modified as circumstances change. The models are proving the French and German philosophers correct as surely as any mathematical proof. And they're beginning to transcend human comprehensibility. Super intelligence is just around the corner, if not here already.

insanitybit 37 minutes ago | parent

This just feels like seeing a complex empirical result and then having to define your view that confirms it. Except in this case the empirical result is a passing proof.

cubefox 31 minutes ago | parent

> Woah, if this proves to be correct it'd be massive, for maths, computer science...and AI too, I guess!

The Riemann theorem (as we apparently now have to call it) is a statement of pure mathematics and has no practical application or interest.

kjanus03 18 minutes ago | parent

It isn't a theorem yet. "No practical application" might have some truth currently but "no interest" is false and laughable..

dropshade77 17 minutes ago | parent

> this feels like we're moving away from the old world of 'science', where we could actually read an experiment, understand its steps, and, with enough time, possibly reproduce it.

It’s a genius first step that we will iterate on and make better.

argentier 16 minutes ago | parent

or there's a bug in the proof checker

which is more likely: that the Riemann Hypothesis is solved, or that a possible solution exhibits a bug in the theorem checker?

demibabs 6 minutes ago | parent

You could use a soundness bug in lean to prove anything. So if there is, it’d be strange that that isn’t happening

Routine8502 1 hour ago | parent

Big if true. Looking forward to seeing it verified!

faangguyindia 1 hour ago | parent

Imagine being an employee at one of the largest AI companies. You can do this:

grep -ri "Riemann Hypothesis" /all-users-chats

Then let an LLM sift through everything, identify potential clues, and pick out promising candidates.

Just think about it: how many people use such LLMs and talk to them about their proofs or research? or their moonshot ideas?

You can have a super-high IQ, but you may not find anything worthwhile, yet a guy less intelligent than you may discover something extraordinary.

To capture that, you need a numbers game. One among billions of dumb people has a higher likelihood than one among 1,000 geniuses.

And LLM companies have all geniuses and ordinary people contributing research for free.

It would be foolish to assume, that people working for LLM companies are coming up with proof all by themselves.

shostack 51 minutes ago | parent

I wish more were shared on how deidentified and anonymized user session data is used when model training is off.

There is a big murky gray area around synthetic data, where and how it is generated, and where and how it may be used for things like this, or model training (because if it is de-identified it is no longer "your" data) or ads measurement and targeting. More transparency is needed there.

staplehorse6553 36 minutes ago | parent

It's almost like everyone was told not to submit all their e-mail (at first) to Google, but hey, who wants sysadmin job openings, amirite, also they can do no evil

jorisw 1 hour ago | parent

For the laymen, what's the significance?

criddell 57 minutes ago | parent

This is the only information I've found that was at all comprehendible by me:

https://en.wikipedia.org/wiki/Riemann_hypothesis#Consequence...

ReptileMan 55 minutes ago | parent

Humongous

rhdunn 55 minutes ago | parent

The Riemann Hypothesis is used as a foundation for several important theorems. See https://en.wikipedia.org/wiki/Riemann_hypothesis#Consequence... for details.

vuldin 55 minutes ago | parent

I have the same question. So far I'm not seeing details on why this is worth continuing to read about.

Betelbuddy 54 minutes ago | parent

Mathematicians are nothing more than a Matrix product.

shawabawa3 53 minutes ago | parent

Thousands of mathematical papers build on top of it, basically proving X holds _if_ the Generalized Riemann hypothesis holds

So all of those X are now proven (if this is indeed proven)

darkwi11ow 53 minutes ago | parent

The direct impact is mostly theoretical, as many things in number theory are based on assumption of Riemann hypothesis, and they were not advancing much historically because there was no proof of Riemann. These things will start getting more spotlight if proof holds true, and they could lead to advancements in cryptography, coding theory, computational theory, and other fields that are heavily invested in number theory.

NB: I am a student of mathematics, not a researcher, and some practicing researcher could correct me here

jrflo 52 minutes ago | parent

It's one of, if not the biggest unsolved problems in math. It's also a millennium prize problem. There are a ton of conditional proofs in math that are true if RH is true, so this would fully prove tons of work. There's also a link between the spacing of nontrivial zeros in RH and the spacing of prime numbers, so it will be interesting to see if this proof illuminates anything about the primes.

TehCorwiz 52 minutes ago | parent

As a fellow layman my understanding is that if it holds true then it has implications about the distribution of prime numbers. Cryptography is an example of a field that relies on prime numbers. They underpin a lot of security. Also, right now finding new large prime numbers is relatively difficult to do, but relatively easy to check. I'm not sure if it's NP-something or not, but finding a solution to shorten the effort to find new primes might have broader applications in algorithm design.

__lm__ 52 minutes ago | parent

It is one of the greatest open problems in Mathematics which has stayed open for more than 100 years (actually since 1859), which has eluded generations of mathematicians, including some of the most brilliant minds of the XX Century (it was already important enough that it is part of one of problems in the list of the 23 Hilbert's problems (stated in 1900). Many results holds under the assumption that RH is true.

nickcw 51 minutes ago | parent

Lots of existing maths proofs start by assuming the Riemann Hypothesis true. Proving it will automatically complete those proofs.

More discussion: https://mathoverflow.net/questions/17209/consequences-of-the...

jtumbot 50 minutes ago | parent

Complicated topic that I don't fully understand, but I believe our understanding of Complex numbers would be much better. In that we understand regular/irregular distributions. The knock on affect is there's thousands of theorems that have been prove 'if' the Riemann Hypothesis is true. And there'll be much more to look into now that it's proven.

I don't know how immediate it would be but you'd expect to see improvements in any algorithm that uses prime numbers. Which is quite a few (RSA, Hash tables, RNGs)

fallingbananna 50 minutes ago | parent

Looking at the wikipedia page explaining it... I get the feeling there is no *quick* way to explain it to laymen.

(Just a feeling, I don't know the significance either.)

keybored 41 minutes ago | parent

I haven’t seen a Wikipedia page on mathematics that was written for laymen yet.

cgearhart 43 minutes ago | parent

In truly layman's terms, this is among the most famous problems in mathematics; it is arguably the holy grail of pure mathematics and number theory. It is also another one of the Clay Millennium Prize problems.

In the ~150 years since it was posed, it has become a lynchpin for hundreds of downstream results that assume it is true. This proof would immediately confirm all of those results.

More practically, there are also some algorithms that will now have guaranteed theoretical bounds and it will likely eventually lead to some more interesting algorithms in the future.

It's kinda the math equivalent of "we found life on another planet".

ryandrake 20 minutes ago | parent

I went to the Wikipedia page for it, hoping to find one of those “practical applications” section and left greatly disappointed! No doubt that this is very exciting for Math departments and would be a historic achievement within the field, but I still struggle to see what real world advances this could bring about. Is this about cryptography or something?

hbcdbff 12 minutes ago | parent

There’s a reason it’s a pure mathematics problem and not applied.

trio8453 41 minutes ago | parent

It's the hardest problem in mathematics and solving it would be the most impressive thing done by AI so far. It also automatically proves many other statements in mathematics which are currently formulated is [if the Riemann hypothesis is true, then <this> happens].

Take everything that happened a couple of months ago with Navier-Stokes and increase the magnitude several times.

meowface 30 minutes ago | parent

Is it the hardest problem, or just the biggest problem? Is solving P=NP not more difficult?

BigTTYGothGF 41 minutes ago | parent

For certain parts of pure math, it would be tremendous, it's been a big deal and is connected to lots of other things in those domains. For everything else in the universe, there's no real consequences.

pg_bot 40 minutes ago | parent

Arguably the most famous unsolved math problem.

charlieyu1 39 minutes ago | parent

The Riemann zeta function can be written as a product of the same function in prime powers.

Finding zeros of a complex function tells us a lot about Riemann zeta function, and in turn, the distribution of primes.

I’d compare this to Fundamental Theorem of Algebra in terms of importance.

est 24 minutes ago | parent

Goldbach's conjecture would turn out to be true.

https://mrc.sdu.edu.cn/english/info/1018/1514.htm

curt15 1 hour ago | parent

It surely can't hurt that this guy is already an expert in the field. How else would one even know where to begin, let alone formulate all the prompts?

darkamaul 1 hour ago | parent

> Claude and I have had a great time together with such breadth in recent months, but this time Claude far outpaced me near the finish line. No, Claude flew, and danced, and sparkled near that finish line.

I quite love the remark here, this feels quite poetic. And it shows, what most experts have already said, that the correct way to use AI to do great things is the Centaur model: frontier AI + domain expert.

howunfortunate 35 minutes ago | parent

For now. Hopefully that holds for a long time, as the alternative is very scary.

blondie9x 59 minutes ago | parent

It’s not a bubble. It’s not a bubble. Don’t worry it’s not consuming resources. The data center push back isn’t real. The public is actually receptive to more AI being involved in their lives.

This is the right direction.

But what if it isn’t?

IndeanCondor 13 minutes ago | parent

Do you believe the Telecom and Dot Com bubbles were fake because fiber optics, lasers and the internet are incredibly useful technologies?

The bubble case is financial. Semicon fabs material input companies have already placed in hiring freezes and stopped all expansion. Wall Street estimates want Nvidia at 17 GW of compute produced next year. Nvidia targeted and met a 10GW production this year, but only 6GW data centre capacity came online in 2026. We are already in overproduction.

The anti bubble case needs to argue that there will be enough demand to want 17GW of compute, enough data centre capacity operational by April 2028 to house that compute, enough energy to power that compute (applying a potential factor of 0.15 for solar, 0.66 for thermal, that's 180GW of solar added in the US or 40GW of thermal for just the GPUs; and data centers need to power much more than just the GPUs, the US added 34GW of solar and -1GW of thermal)

With these numbers, we could have the singularity tomorrow and it still wouldn't be possible to physically move the equipment and material and generation to turn all of this on in time to meet the estimate targets that the industry is operating on.

albinstman 57 minutes ago | parent

Interesting timing, given that the claim comes just a day after Anthropic’s IPO.

lamename 49 minutes ago | parent

The author addresses this in the document. Not that strongly in my opinion, but at least addresses it.

lou1306 39 minutes ago | parent

If you look at the timeline, the last line says "11/16/26: GRH". Thus it seems Claude got the proof four days _before_ the IPO.

fwlr 38 minutes ago | parent

This is discussed in the pdf, where he says he had the proof finished “on [a] Sunday” (i.e. before the IPO), and chose to wait until after it to announce.

It is prudent to still reserve some suspicion given the immense wealth at stake, but if this is substantially true, it is undeniably admirable.

port3000 54 minutes ago | parent

Thank god it was proven and not disproven!

nisegami 40 minutes ago | parent

Definitely, it would have genuinely been upsetting if disproven.

jackschultz 54 minutes ago | parent

> It is natural to feel that this is an end, in my case a loss of a principal guiding light, but what I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater.

> Claude and I have had a great time together with such breadth in recent months, but this time Claude far outpaced me near the finish line. No, Claude flew, and danced, and sparkled near that finish line. I had us on the line of combining two quite different ideas present in a toy setting, one of which was already new and shocking, making very simple one of the shadows of Riemann over which we mathematicians felt most triumph. I guided us to the right situation analogous to that toy setting, and then Claude began to leave me in awe.

> But I hope in having the experience of seeing all of number theory’s beauty brought together I am just one of the first, and those who love the mathematics will all be able to enjoy such transcendence in the future.

Bus meme irl. I guess I'm lucky that I have this experience even when working on boring things like data work. But these types of attitudes can really make a difference in how we feel and our end results.

FartyMcFarter 53 minutes ago | parent

At some point I was looking at ways to try and find counter-examples computationally. I'm glad I didn't spend too much time on that...

xbryanx 53 minutes ago | parent

For the non-mathematicians and folks interested in real-world applications, what is the impact of the Riemann hypothesis being proven? Are there major breakthroughs in cryptography or other areas of computer science or physical science that flow from proof of this hypothesis that was generally already considered true?

rf15 36 minutes ago | parent

it's largely confirming what we already assumed to be true, so nothing much changed, apart from certainty. (If this thing is correct anyway)

diomedes 27 minutes ago | parent

not much will change other than knowing that some forms of cryptography are more secure. if it had been false, things would have changed much more drastically

cubefox 22 minutes ago | parent

No, all the millennium prize problems (and pure math theorems in general) are without any practical value.

simianwords 52 minutes ago | parent

It’s funny that this happened after the IPO and not before. I’m sure he timed it that way

nisegami 39 minutes ago | parent

He talks about it in the paper and is quite clear that this was not his intention. Whether you believe him or not is a different story, but personally I do.

Hamuko 23 minutes ago | parent

Well, it sure isn't helping Anthropic's stock. ANT closed yesterday at $97.17, opened today at $89.86 and is now trading at $86.00.

reworkal 51 minutes ago | parent

> No, Claude flew, and danced, and sparkled near that finish line.

Side note, anthropomorphizing your stochastic parrot sort of distracts me from your argument. Like sure guy might be a genius but as someone who knows nothing about you, reading this is very off-putting.

This is like my own opinion.. not saying what he wrote is bad, hope its real TBH

timmmmmmay 44 minutes ago | parent

how many unsolved math problems does it have to solve before you drop the parrot thing

luka598 28 minutes ago | parent

well, if you make the parrot big enough and teach it enough skills it perhaps becomes usefull, but it still is just a parrot

azan_ 21 minutes ago | parent

Yes, if you stretch your definition far enough everything will be a parrot, including humans.

chpatrick 18 minutes ago | parent

What proof of the Riemann hypothesis is it parroting exactly?

cman1444 9 minutes ago | parent

I guess we're all parrots then.

munchler 44 minutes ago | parent

It might be off-putting, but perhaps it’s time for you to face the fact that Claude (and other high-end models) are way past being stochastic parrots.

chilipepperhott 5 minutes ago | parent

I think you're missing his argument. Claude is a tool, not a person. It's just as counterproductive to anthropomorphize my hammer as it is to anthropomorphize an LLM.

azan_ 42 minutes ago | parent

> stochastic parrot

This has to be satire, right?

dude250711 29 minutes ago | parent

Could be a time traveller from 2024.

jimmysnuka 29 minutes ago | parent

Can you provide a convincing argument that your brain is not also a "stochastic parrot"?

kyle-rb 12 minutes ago | parent

He said "Claude flew", which fits perfectly with your stochastic parrot framing. He's not anthropomorphizing, at worst he's ornithomorphizing(?)

elffjs 51 minutes ago | parent

Alpöge is moreover claiming the Generalized Riemann Hypothesis. The written argument is over 250,000 pages long and not provided.

The toolbox here contains a surprising, to me, amount of algebraic stuff.

imhoguy 39 minutes ago | parent

CS layman folk here. How one is going to prove 250K page argument? Just single sign of hallucination can make it false, no?

tasuki 29 minutes ago | parent

NoDodgeQuestion 27 minutes ago | parent

AI proof in lean, can verify by computer. Mistake can only be compiler bug or state problem wrong.

well_ackshually 3 minutes ago | parent

Mathematicians, famously known for being extremely precise in their specification of problems and absolute lack of errors writing down proofs.

It's always, always, always stating problem wrong.

Epa095 25 minutes ago | parent

Its formalized in lean, which verifies each step. The steps themselves are relatively easy, and can be mechanically verified (think "A and A->B thus by modus ponens B").

Traditionally writing verified proofs is a hard undertaking, exactly because you need to write out a lot of steps "normal human" mathematics jump over, but with AI its easier.

vintermann 19 minutes ago | parent

At what point do we start worrying that it's found a bug in Lean instead?

est 12 minutes ago | parent

> The steps themselves are relatively easy, and can be mechanically verified

They aren't

cubefox 23 minutes ago | parent

Small errors in proofs can often be fixed easily. So it's not that the whole thing would be useless. Also, he has a Lean proof, which rules out these small errors.

adgjlsfhk1 20 minutes ago | parent

Step 1 will be making a much shorter argument.

tasuki 30 minutes ago | parent

It is linked from the pdf, just rather big.

layer8 20 minutes ago | parent

The Lean proof is linked, not the 250,000-page writeup (because it wouldn't be very useful in that form).

fantasizr 20 minutes ago | parent

It is sorta amusing that you're asking the community to review 250,000 pages of AI generated work. Could all be true, but that's a big ask of a human. And asking another AI to review it seems something he could just do himself.

synapsomorphy 50 minutes ago | parent

That was fast. I thought we’d have more time.

I wonder if we’ll have room temp superconductors and fusion this time next year.

If this holds up it seems as good a day as any to call the end of one time and beginning of another.

I wish I could be fully happy about this new era instead of worrying if I’ll be alive in 5 years or not. If there was any time for Anthropic to try to coordinate a pause, it feels like now or never, they seem to have a pretty strong lead at the moment. But probably they won’t and can’t, and we will just put the fate of our species to a coin flip.

karmakurtisaani 44 minutes ago | parent

> I wonder if we’ll have room temp superconductors and fusion this time next year.

Still waiting for the AI advancements which are not just taking away work from SWEs and mathematicians..

ndriscoll 35 minutes ago | parent

Why? Even as a SWE there's plenty of software that I want in my own life that I simply don't have the time/attention to make, or even to try to commission someone else to make (e.g. I've been making little educational games/demos for my kids). Having something that does what I want in a few seconds when I have the attention is fantastic.

"Taking away work from SWEs" means enabling everyone else (or even SWEs) to do things they couldn't do before. This is obviously an improvement.

vidarh 20 minutes ago | parent

I have a todo list long enough I don't even have time to prompt for all of it...

If I could justify enough tokens, I'd point an agent at my harddrive and tell it to formalise all of my old notes and goals, and spin up agents to build all of the things I've made notes about that I would like to exist. It'd still likely take years...

I just kicked off Kimi building a game based on a concept I had kicking around. I wouldn't have time to build the game or even write the spec, but I had time to type "write a game based on [paste of the concept]" to see what falls out of it.

I can't imagine ever running out of things I want, because every thing I finish tends to result in my todo list growing with new things.

xvector 32 minutes ago | parent

I truly don't understand how someone can be interested in technology and visit HN without understanding - even broadly - the history of technological progress and how these things take time.

Yeah, I want the cure for cancer yesterday as well, who doesn't? This is the most significant result in mathematics in over a century.

NitpickLawyer 30 minutes ago | parent

> Still waiting for the AI advancements

At this point this line of "yeah but what about..." is getting silly. We've had 2 millennium prize problems fall this year, plus a lean proof for FLT (another effort worth 1m$) and people are still "yeah, but what'd be cool is if..."

Dunno, at this point I feel like people are trolling. This is absolutely insane pace, and just last year we were hearing how "there is no more data", "models are stalling" and so on. The negativity and goalpost moving is getting tiring.

vintermann 21 minutes ago | parent

Surely faster than light travel is worth more than a million dollars.

(Yes, yes, I know it's Fermat's Last Theorem)

shigawire 9 minutes ago | parent

I think it is because we see these headlines, but as a layperson it has no impact on my life.

I don't see the value and I don't see the world changed.... yet

tempest_ 14 minutes ago | parent

They are already there. Sure you are not seeing positions disappear but hiring is starting to grind to a halt in a lot of places. Someone who might have needed an admin assistant no longer needs one and middle management in large orgs whose job was basically summarize and pass up will soon start to thin out through attrition.

You wont see whole sale replacement but you will see doing more with less.

xvector 36 minutes ago | parent

They definitely don't have a strong lead, trading places with OpenAI and occasionally Google every few weeks

A coordinated pause needs to come at the national level

WarmWash 26 minutes ago | parent

We're in a cold war

inkysigma 32 minutes ago | parent

I do wonder how well a lot of this translates into physical sciences. It's one thing to be able to RL on lean but physical sciences don't have such an obvious quick feedback loop. A lot of the scientific results have been closer to the theory side which isn't unexpected from the sheer mathematical capability.

teravor 18 minutes ago | parent

if you look at the current "high" temperature superconductors it becomes clear that a peculiar atomic arrangement is the key yet we don't have the ability to individually place atoms (yet) so we must rely on entropy to place them into said arrangement for us which restricts the space we can even test.

so the task for an LLM here is doubly difficult and the feedback loop is of very high latency. I do wonder if someone is trying though. simulate physical process and entropy/annealing -> simulate superconductivity. and very few working examples to work with.

inkysigma 8 minutes ago | parent

So a couple of purely informational question coming from a layperson:

1. Is there a good theory of high temperature super conducting that we can easily simulate? My furthest understanding of this was that BCS was insufficient.

2. Does such a theory impose limits on the temperature that are achievable?

3. Do we know if such a theory is complete in some sense? (e.g. if such a theory predicted no super conducting material at a temp K and pressure P, we would know that there could be no such material at temp K and pressure P based on reasonable assumptions)

Duwensatzaj 25 minutes ago | parent

We will not. Physics experiments take time and money to complete.

geraneum 24 minutes ago | parent

> I wonder if we’ll have room temp superconductors and fusion this time next year.

That’s child’s play. I wonder if we’ll have star trek replicator by next year.

cbg0 20 minutes ago | parent

> fusion this time next year.

Aren't most of the remaining problems related to nuclear fusion engineering ones? Like building reactor walls that survive neutron damage.

ariwilson 50 minutes ago | parent

The way this is getting announced via a human mathematician humbly presenting a result is great, as opposed to the marketing influenced OpenAI post.

I am still a bit sad though. I had hoped the Rieman hypothesis would fall mostly to a human mathematician rather than 60/40 AI/mathematician that my layman's read of that note interprets.

alex43578 44 minutes ago | parent

I don’t get this feeling. Landing on the moon isn’t less of an accomplishment because we built the Saturn V instead of walking. If anything, can’t we appreciate how we took some sand and made it smart enough to help us solve problems?

abixb 38 minutes ago | parent

Yes. We're tool using species. We evolved from African savannah to this AI moment by using tools.

I don't understand this skepticism towards tool use. It's like saying, "I'm sad that Reimann Hypothesis solution was published as a PDF and not in a parchment or made as a verbal announcement."

Anyway, great accomplishment. What an incredible thing to wake up to.

psychoslave 30 minutes ago | parent

Now that you suggest it, a video of the person declaiming the announcement while holding a parchment, would definitely would be nice to have. Hopefully someone will generate that using some genAI video tool.

ariwilson 37 minutes ago | parent

I also build things that would've taken me months in days now so I do get that feeling of amazement at what is possible now.

But at the same time one of my favorite memories from childhood as an introvert nerd math / computer guy is locking myself in a room for an evening and relying on my brain to solve and truly understand a problem. I'm nostalgic for that. A solution that requires 250k pages to describe takes something away from that.

Humans are complicated and contradictory, which is great!

jm4 31 minutes ago | parent

The entirety of human history is about building tools to perform tasks our bodies aren’t capable of on their own. Someone else in this thread said something about how cool it is that humans turned sand into a machine that could accomplish this. Our gift as humans is that we can devise ways to overcome the limits of our brains.

peteey 34 minutes ago | parent

A more apt comparison is getting to the moon by outsourcing the Saturn V design to China.

Previously our talented contribution was doing the thinking. Now, AI seems to do this as a separate entity.

tock 32 minutes ago | parent

The moon landing was an achievement for all of humanity.

throwaway132448 15 minutes ago | parent

But the Nazis in particular.

SoftTalker 32 minutes ago | parent

I can lift more weight with a forklift than I can on a barbell but it doesn't give me any sense of accomplishment.

pembrook 30 minutes ago | parent

If you built the forklift you'd have a sense of accomplishment.

acedTrex 25 minutes ago | parent

Alpoge did not build the forklift, kind of a basic flaw in that analogy.

ben_w 27 minutes ago | parent

Aye there's that, but I'd feel pretty accomplished if I built a working forklift.

Or used a forklift to build a house that everyone else said couldn't be built because it was beyond human muscle.

WarmWash 30 minutes ago | parent

We are all sitting on a thousand years of media telling us that the human mind is exceptional and transcendent.

It's naturally going to be upsetting to everyone if reality unfolds to show that human intelligence is nothing special, and actually not even that good.

NoDodgeQuestion 26 minutes ago | parent

special yes, good no

layer8 30 minutes ago | parent

It would be sad if AI designed and built the Saturn V and we had it fly us to the moon without understanding the details of of how it is built and how it works. There is furthermore value for humans in going through the discovery process themselves, not just value in getting to the moon. So I understand how one can be sad if that won't happen anymore now, and instead humans are relegated to analyzing what the AI designed and built.

ChicagoBoy11 18 minutes ago | parent

But to extender your analogy: We do understand how Claude does what it does and how it is built, so how is this different?

barbazoo 12 minutes ago | parent

And presumably mathematicians understand the output, the proof, whatever it is that came out of the process. I don't feel like humans have to be the one doing all the manual math and figuring out all combinations.

layer8 11 minutes ago | parent

I would dispute that we completely understand how LLMs do what they do. An analogy would be that we meet an alien race that gives us their proof of the Riemann hypothesis translated to Lean. Plus of all other provable mathematical conjectures we come up with. And we might kind-of understand how those aliens do math. It would still be a bit sad that we'd be relegated to consuming what the aliens give us, and never be able to ascend to the level of the alien mathematicians

Philpax 9 minutes ago | parent

We definitely don't understand how Claude does what it does! We know how to grow a Claude, but we don't understand what goes on inside a Claude.

streetfighter64 7 minutes ago | parent

I mean I understand how cars work but that doesn't mean I'd feel the same accomplishment of driving 42 km as running a marathon.

uludag 30 minutes ago | parent

I think a more apt analogy which gets at what they probably felt was immagine if humanity accomplished the moon-landing via a supposed Walmart Inc. Space Division, the announcement of celebration of this event felt like a giant Walmart promo, and the only reason they wanted to go to the moon was to put ads on it.

all3z 3 minutes ago | parent

You mean like that energy drink company that has a better space program than most countries?

Flatterer3544 29 minutes ago | parent

Some things getting solved opens a new whole dimension of new possibilities, so assumebly the negativity steems towards the fact, if this did just take us from A to B and there were no new undiscovered paths encountered during the journey to be explored..

acedTrex 27 minutes ago | parent

I can not fathom this viewpoint, it is as alien to me as being a flat earther. What frame of mind is required to consider an analogy of "we built saturn V instead of walking" to be even remotely ontologically in the same stratosphere as what is happening today?

I can't even begin to make sense of that lack of basic understanding of humanity.

ebb_earl_co 25 minutes ago | parent

It seems to me, having B.S. and M.S. in mathematics, that the intrigue of human intellect continues to dim. I would wager that a large cohort of mathematicians, even ones remote to the discipline now such as myself, maintain the joy of cogitation.

One felt a camaraderie with all mathematicians in history because mathematics was, wonderfully, a club with no bouncer, the only cover charge being a bit of concentration of attention. Not that one could join the elite ranks of Gauss, Euler, et al. by association, but the bonhomie was real for a lot of us, realizing that all of the legendary mathematics was dreamt up in the same way that we were engaging in it, too.

That horse has left the barn, decidedly, and there is grief.

curt15 19 minutes ago | parent

The effect of LLMs on math is pretty much the same as for SWE: great for existing experts like this guy, bad for juniors since the usual criteria for evaluating merit are no longer useful, and completely demoralizing for prospectives.

teeray 21 minutes ago | parent

It’s a bit more like we created sentient animals in a lab, they ran off to another planet, then sent one of their spacecraft back to shuttle us to Ganymede. We have no idea how the spaceship works, and if it never arrived, getting to Ganymede would still be science fiction. You can’t authentically call that an accomplishment… just a result really. We get a participation trophy at best.

xtracto 19 minutes ago | parent

For the first time in humans 200,000+ years we have an "independent" or "different" intelligence that can teach us things... and WE built it. We can learn from it and improve after its teachings. This new era is going to be amazing.

iknownthing 43 minutes ago | parent

Don't understand the sadness - are you interested in the discovery or credit attribution?

whycombilater 21 minutes ago | parent

People often prefer horses to combine harvesters.

Zambyte 18 minutes ago | parent

The process of solving hard problems like this often involves inventing / discovering new branches of mathematics in order to solve the problem. By solving the problem with AI, we haven't potentially snuffed out the potential for discovering these new branches of mathematics.

bitwize 43 minutes ago | parent

Get used to it. No human will prove anything significant unassisted, ever again. This is reality now.

psychoslave 30 minutes ago | parent

Can you prove it?

nozzlegear 30 minutes ago | parent

Get a load of Nostradamus over here

parineum 24 minutes ago | parent

No human has proved anything significant "unassisted" in quite a while.

dahart 23 minutes ago | parent

This kind of assistance is new, but I’d say nobody has ever proven anything truly unassisted, and people have been trying to point out for a long time that crediting a single person for discoveries and claiming they were “unassisted” conceals the true story. Education, background knowledge, conversations, correspondence, and context have always been paramount. This is one reason Math has often hit points where multiple people make the same discovery at nearly the same time “independently”. It might be a good thing that we can now dispense with the myth of anything being “unassisted”.

bitwize 18 minutes ago | parent

I mean unassisted by AI.

Waterluvian 37 minutes ago | parent

Real mathematicians do all their work with calcium carbonate, not silicon dioxide.

marcusverus 35 minutes ago | parent

The F-47 is pretty sweet, but what a shame that it was made without employing a single, solitary slide rule!

mahogany 29 minutes ago | parent

Probably closer to 90/10

suddenlybananas 13 minutes ago | parent

None of us can know this without all the prompts/history

Sol- 23 minutes ago | parent

I think Alpöge specifically was criticized in the past for being somewhat disrespectful of the math community, since he just dumped AI slop onto Twitter in meme form. The OpenAI post was actually more respectful compared to that. So this type of announcement is more of a return to somewhat more deferential communication.

dist-epoch 22 minutes ago | parent

40 percent human? In this case the human has 0 percent math input.

It might have a lot of engineering credit (setting up the swarm, ...)

suddenlybananas 12 minutes ago | parent

You didn't read the PDF, that's not what Alpöge did.

dist-epoch 10 minutes ago | parent

I did read it:

> I'm just too inexpert in the relevant stuff at the moment, I hope the reader sees the breadth of this argument!

> I intend to do my best to understand and impart understanding of this argument

He was shepherding Claude.

suddenlybananas 4 minutes ago | parent

> I guided us to the right situation > I had us on the line of combining two quite different ideas present > The parts in which my vague intuition worked out due to Claude’s technical mastery

I am not saying he could have done this without Claude. But there's a reason he did it and not anyone else. This wasn't a case of giving Claude the prompt "solve riemann hypothesis make no mistakes"

rustyhancock 9 minutes ago | parent

Given Anthropic IPO was yesterday, I suspect anthropic needs to be cautious about how they present anything not entirely confirmed.

gpm 45 minutes ago | parent

Note that the pdf contains links to the lean formalization

https://alpo.ge/lean.tar.gz.part01 through https://alpo.ge/lean.tar.gz.part14

falcor84 8 minutes ago | parent

I suppose we need a new online platform to properly review large lean proofs.

varispeed 45 minutes ago | parent

That is great, but why Anthropic and other companies are allowed to hoard and gatekeep resources? Businesses and consumers already struggle with buying computing at prices they can afford.

To me this reads like: "I bought all your water and solved my thirst, peasants!"

rhdunn 45 minutes ago | parent

> once we saw multiplicity 2 on O(2, n) via the "natively-automorphic" determinant ... meaning holo/automorphy over a function field

So if I understand correctly this proof has proofs for additional things that are steps to the main proof. That would be similar to the proof of Fermat's Last Theorem that required proving the Taniyama–Shimura conjecture [1].

[1] https://en.wikipedia.org/wiki/Fermat's_Last_Theorem#Subseque...

JumpCrisscross 39 minutes ago | parent

> I had a stream-of-consciousness jaunt

What? Why? Did nobody—not even an agent—copy edit this? Why is the first half a Turgenev novel?

fourseventy 39 minutes ago | parent

we were 2 days shy of being able to say "we got reimann hypothesis before GTA6"

rf15 37 minutes ago | parent

> The argument is incredibly long at the moment

so... we have plenty of places to hide logical flaws, got it

cgearhart 37 minutes ago | parent

This is such a wild ride. The conjecture itself is not so hard to understand, but approximately _all_ of the maths here are beyond me. The paper is written up to suggest "we knew which direction to go, but really had no idea how far the journey would be" which certainly makes it seem like the human expert was useful in this process. Perhaps that's just wishful thinking, but it's consistent with what I see elsewhere that AI is having an effect. The biggest effects accrue to the most effective humans. Random folks prompting is better than random folks toiling alone, but experts prompting is a different thing.

mellosouls 36 minutes ago | parent

From the link:

It is natural to feel that this is an end, in my case a loss of a principal guiding light, but what I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater.

If the proof holds up, I suspect that will become a classic quote.

yewenjie 34 minutes ago | parent

I know this thread will be full of people in denial about full recursive self-improvement / intelligence explosion / superintelligence (the original meaning of the word) within months or a year, so just quoting Scott Aaronson:

"Experience has shown that, even now, there will still be people explaining in patronizing tones why none of this is real and none of it counts. If such people were capable of being impressed by anything that happens in the empirical world, of updating on anything, they would’ve already been impressed and already updated several years ago, long before things had reached the point of an actual Mathocalypse."

geraneum 28 minutes ago | parent

> superintelligence within months or a year

Didn’t we already achieve super intelligence?

wslh 26 minutes ago | parent

I would say, seriously, that this is bounded superintelligence. That superintelligence needed top human talent to achieve the goal. It was not a prompt like: "solve the Riemann hypothesis".

Coincidentally, yesterday I did a small presentation on cybersecurity and AI where I made a similar point: scarce top talent in cybersecurity (e.g. Calif) gets superpowers with AI, while AI replaces a lot of the skills of average cybersecurity professionals using simple prompts.

I think the breaking point is when AI can solve complex problems on its own that even top talent can't achieve. IMHO, I've seen some of this in the latest Google models (especially Carbon, and then Argon), where they can do more research independently.

layer8 24 minutes ago | parent

"Human-assisted superintelligence" (HASI). ;)

jrflo 25 minutes ago | parent

I'm always shocked by the number of smart, technical people who seem to think that AI is going to crash and burn and go away. In 2025 AI got a gold at the IMO for the first time, a competition for high school kids. Now it's likely solved the biggest problem in math. There's no reason to expect progress to stop here, especially on domains highly amenable to RL.

throwaway384034 22 minutes ago | parent

Level of intelligence doesn’t matter if it is not applied to truth searching and instead to reasoning backwards from personal wishes.

cman1444 17 minutes ago | parent

Isn't this post an example of the former?

mvkel 4 minutes ago | parent

Define truth

lordnacho 18 minutes ago | parent

I actually asked a guy who got an IMO gold what he thought about it, and he's happy to concede that AI works. He used to be my go-to guy for those kinds of problems when I came across them. Now you can actually get it to solve those things on your standard plan.

vagab0nd 12 minutes ago | parent

"If you are smart, why can't you predict the future?"

jrflo 9 minutes ago | parent

It's not an ability to predict the future, but the reluctance to step back and look at the trendline rather than saying "But it still can't do X, Y or Z!".

rfgplk 9 minutes ago | parent

Because they're _desperate_ to keep their social standing. That's literally all there is to it. The same occurred during Enlightenment and the industrial revolution. A small upper echelon of society (the church or aristocrats respectively) were deemed worthy enough to possess knowledge and thus to belong to an upper strata. The same story can be seen pretty much in every civilization btw (Greeks/India/China/Japan etc). When the masses started gaining access to this knowledge there was _massive_ pushback, even armed revolutions in certain cases.

What's happening today is that anyone and everyone can have access to superintelligence. Anyone with access to an LLM effectively is, simultaneously, a world-class diagnostician, engineer, architect, musician, mathematician... you get the point. This will necessarily lead to a post-scarcity society, since everyone knows everything and will be capable of producing anything.

bodge5000 5 minutes ago | parent

> This will necessarily lead to a post-scarcity society

I wish I shared this optimism, unfortunately that upper echelon of society is the one which currently controls the superintelligence. Everyone may have access to it, but they own it.

dboreham 24 minutes ago | parent

We're really using a term invented by Trump yesterday for something we've been working on for 80 years?

bananaflag 20 minutes ago | parent

Superintelligence is an old term with an established meaning. What Trump does is to use it for AI in general.

chilmers 16 minutes ago | parent

The term was not invented by Trump. For example, the (in)famous book Superintelligence by Nick Bostrom was published in 2014[1]. It has a long been used to refer to a theoretical general AI that far surpasses humans.

The Trump thing is a insane newspeak attempt to forcibly replace the use of "artificial intelligence" with "super intelligence" in a general sense. It is ridiculous, obviously, but the best thing we can do is to continue using "superintelligence" in its original, narrower sense.

[1] https://en.wikipedia.org/wiki/Superintelligence:_Paths,_Dang...

whycombilater 11 minutes ago | parent

Actually, I think the best thing we can do is take Trump round the back of the woodshed.

codewithcheese 23 minutes ago | parent

RL is not scalable to super intelligence. Super intelligence should be able to learn continuosly without verifiable rewards.

StevenWaterman 15 minutes ago | parent

You don't need RL all the way to ASI. You just need RL far enough that the model becomes good enough at AI research that it discovers something better than RL.

codewithcheese 11 minutes ago | parent

> becomes good enough at AI research that it discovers something better than RL

That might require continuous learning without verifiable rewards. RL models might help humans join the dots faster, but we don't know how many years that will take.

CarlitosHighway 21 minutes ago | parent

Both the deniers and claqueurs are from the same ilk: Silly adults who don't have the intellectual capacity to create or even fully understand the subject, yet possess an almost vulgar self-confidence to constantly opine about its future.

FLeXMurphy 19 minutes ago | parent

Good observation. Perhaps the noble path is to chop wood and carry water, after all.

Hmorgan12 19 minutes ago | parent

The nature of mathematical proofs requires us to stay vigilant and careful. OpenAI pulled some of the research results from their famous repo due to simple mistakes that where easily spotted by human experts. There have been infamous mass of incorrect Fermi's theorem proofs.

LLM-assisted research is proving to be useful, rewarding and joyful, but we should not treat LLMs as a perfect machine, incapable of making mistakes. Especially in such massive proofs as this one.

I won't say that it "doesn't count". I will simply ask for staying rational.

asdfologist 18 minutes ago | parent

But it's not clear how well superhuman ability in math generalizes to other domains.

deprave 18 minutes ago | parent

I think this is all real and counts. Where I struggle is explaining to non-technical people how their everyday lives are going to change directly or indirectly. I hope these discoveries are the first in a sequence that will culminate in or surpass a cure for cancer and diabetes, cheaper and cleaner energy and water, and other tangible everyday life improvements that influence millions of people.

vinyl7 16 minutes ago | parent

A cure for cancer and disease will only be available to the oligarchy, we the peasants will be killed off since we're no longer needed.

prmph 5 minutes ago | parent

[delayed]

toddmorey 9 minutes ago | parent

What is the action item exactly for non-technical people? How do you prepare for “things will be different in unpredictable ways at an unpredictable time”? Even if that time moves sooner, still feels like very little info to truly act on.

If you watch movies or read fiction, we’ve expected for years that life in the future will be very different.

jubilanti 18 minutes ago | parent

> superintelligence

This is the term you use when you intentionally want to align yourself with the Trump regime.

whycombilater 14 minutes ago | parent

....unless you literally do have a new silicon god (in which case Trump is irrelevant).

Tanjreeve 18 minutes ago | parent

It’s a bait and switch to use a harvard fellow of mathematics solving a problem using AI with “The AI is super intelligent”. People are getting really tired of the constant bait and switches and subterfuge in press releases so unsurprisingly they just start rejecting it by default.

cbg0 17 minutes ago | parent

I don't think you need to be in the football team mindset of us vs them, we're all seeing the progress but most of it is still built on top of what humans have put together over decades and not completely new inventions, which is probably what would convince most that we've achieved "superintelligence" or whatever you want to call it.

soon_to_be_sb 16 minutes ago | parent

I am not impressed because I understand how it works. Not because what things it can or cannot do..

We can start talking when we can train these things without having to beat on its head a million times in a million different ways to teach it one little thing...

bodge5000 16 minutes ago | parent

My thoughts remain the same since I've been hearing "months or a year" for the past 5 years. Maybe this time truly is different, but unless acceptance of super-intelligence is required to take some action (most people I've seen who hold this view also don't think it should be seriously regulated so I'm not sure what that would be), why not just wait until it happens rather than making more predictions? The only thing I've ever seen come out of these predictions is willing me to "stop resisting", but there's no reason to do that until its here.

If you do want to see something done about it and this isn't just another prediction or a vague warning, more power to you, but otherwise, I'll wait and see

bananaflag 12 minutes ago | parent

> My thoughts remain the same since I've been hearing "months or a year" for the past 5 years.

No one said that. If you look at e.g. the Metaculus predictions, you can see AGI has been hovering around 2027-2030 for 5 years.

https://www.metaculus.com/questions/3479/date-weakly-general...

https://www.metaculus.com/questions/5121/date-of-artificial-...

bodge5000 8 minutes ago | parent

Will OPs prediction be counted on Metaculus? If not, it's clearly not an accurate source for "No one said that".

igleria 16 minutes ago | parent

> people in denial about superintelligence

As long as I get to keep my house...

FiberBundle 12 minutes ago | parent

When Atari games or Go were solved around 10 years ago, nobody was saying that superintelligence was within months just because reinforcement learning worked on a domain that allowed learning by self-play. Now we have a domain that is solvable by reinforcement learning with verifiable rewards. It's very impressive that these models are capable of doing the things they do, but the techniques aren't transferable to the vast majority of other domains. ML research also isn't verifiable in the sense that math is. People who claim it's verifiable, because there's a target that can directly be optimized don't understand why RL works for math. It's that lean is able to massively shrink the search space. This simply cannot be done in other domains in the same way.

caidehen 5 minutes ago | parent

I won't believe it until I get a robot to put my dishes into the dish washer, put copper meshes and squeeze spray foam under my siding panels.

xpct 3 minutes ago | parent

What's so surprising about this? A lot of these problems can only be solved using a significant amount of capital. At home or work, using a few-hundred dollar token budget, it's mostly a slog and requires handholding for the majority of open-ended problems.

This technology won't be available to the broad public in the scale that it's available to a select group. If it can only be deployed to a limited amount, to most it remains a magical inaccessible box. And that majority should naturally be incentivized to look for cheaper alternatives, and to "keep moving the goalposts", or in simpler terms, looking where they can provide value.

vessenes 32 minutes ago | parent

He casually mentions it also solves the Generalized Riemann Hypothesis and thinks it will take down the Extended Riemann Hypothesis.

Generalized: superset of the Zeta function.

Extended: I'm fuzzy on these details, but I think of it as Riemann over number fields, not just the complex plane.

I am elated to read about this, it's amazing. Sounds like there's some genuinely interesting and new math in there, which is also awesome.

HappyPanacea 16 minutes ago | parent

Also Grand Riemann hypothesis and Riemann hypothesis for Selberg class as even more general.

Also if Grand Riemann Hypothesis is out of reach right now, can we get a quasi-Grand Riemann Hypothesis for Ramanujan tau L-function in the mean time?

hypendev 32 minutes ago | parent

I kind of expected this to happen by EoY.

Combining the brightest mathematicians we have with intelligence-in-a-box, it was just a matter of time before the hardest math challenges fall.

Congratulations to Levant, but also to us as a species. We have managed to create machines made out of pure math that contain insurmountable knowledge and are helping us solve the mysteries of the universe.

loubbrad 32 minutes ago | parent

It's been almost exactly a year since GPT-5 first started solving Erdős problems. What a time to be alive!

buzzy_hacker 32 minutes ago | parent

If true, it seems like Goldbach's conjecture may fall shortly after, see this recent proof conditional on Generalized Riemann Hypothesis [0]. It has other conditions too though.

[0] https://mrc.sdu.edu.cn/english/info/1018/1514.htm

ogogmad 31 minutes ago | parent

GRH is a juggernaut. Nay, a keystone. Now number theory can get tidier and more user-friendly.

amir734jj 31 minutes ago | parent

If Reimann hypothesis is true, then Goldbach conjecture is true as well. It's major.

kittikitti 31 minutes ago | parent

One of my recent, fully human, mathematical discoveries relies on the Reimann Hypothesis. It's likely to still get dismissed because I don't have a gatekeeper backing me up, but I can rest easier knowing this. I know I haven't been nice to Anthropic but I truly appreciate this. Now all the haters will have to find something else to trivialize about my work!

yodsanklai 30 minutes ago | parent

> The argument is incredibly long at the moment, and while its formalization is finished and it is routine to produce a huge, unreadable document containing sufficient detail, in my view of my path I have just taken a step. I intend to do my best to understand and impart understanding of this argument and its certain simplifications, to speak about it, to write it properly.

Next step, make the proof intelligible for humans.

falcor84 11 minutes ago | parent

I'm afraid that it may have irreducible complexity beyond our naturally evolved cognitive mechanism. And with my sci-fi helmet on, I'd posit that the next step might actually be to cybernetically augment the human mind to the extent that we'll be able to grok it.

Hmorgan12 28 minutes ago | parent

Wow this can take years to verify, maybe even decades

layer8 16 minutes ago | parent

You only have to verify that the Lean statement that was proven correctly represents the Riemann hypothesis, and that the Lean implementation has no applicable bugs.

claaams 28 minutes ago | parent

I wonder if the money open ai and anthropic have spent just on solving these problems was spent on human mathematicians and resources would we have gotten these solved? Sooner?

azan_ 25 minutes ago | parent

Remember that solving these problems is just side-gig for AI companies. So the comparison is not really fair.

neuwirth 26 minutes ago | parent

The shift now needs to be towards statements that can't be posed as a singular Lean statement (or, even, a collection of Lean or similar statements that is understandable by a human). Mathlib/NumberTheory/LSeries/RiemannZeta.lean already had the statement formalized in a way that us mortals can understand. The great open problems of the future will be the ones which cannot so easily be interpretably formalized in the first place, I fear.

lynndotpy 26 minutes ago | parent

Going to be hard getting used to saying "the Riemann Theory".

num42 26 minutes ago | parent

If this is true, it would be huge. OpenAI would race desperately to tackle the remaining Millennium Prize Problems. Perhaps they would even turn their attention to solving P vs. NP with unprecedented urgency.

dooglius 24 minutes ago | parent

David Hilbert:

> If I were to awaken after having slept for a thousand years, my first question would be: Has the Riemann hypothesis been proven?

empath75 23 minutes ago | parent

We almost got a proof of the Riemann Theorem before GTA VI

andyjohnson0 21 minutes ago | parent

This is currently #1 on the home page. And there's apparently a 250,000 page Lean proof.

And #2 is Feynman's saying that "If you can’t explain something in simple terms, you don’t understand it".

So: is anyone going to be able to explain or understand this? Or is that the old paradigm now?

CharlieItanium 15 minutes ago | parent

The general public doesn't read proofs. They're CS graduates cosplaying as mathematicians and just assume the proof to be true and useful. They don't understand and they don't care. I'd like to see some of these people back in their basic engineering classes, asked to prove Shannon's Sampling theorem.

machina_ex_deus 20 minutes ago | parent

The real danger is just around the corner, and it has nothing to do with superintelligence: if these models are so good at math, they might break modern cryptography. All it takes for massive worldwide chaos is these models solving elliptic curve cryptography. Superintelligence or not, if these models can solve cryptographically relevant math questions, faster than we can invent new cryptographic protocols, everything we rely on digitally can break. Imagine the internet without encryption. That includes credit cards, operating systems secure updates in a time where models find vulnerabilities every day, etc. I would start pushing post-quantum cryptography protocols not because of quantum computers, but because potentially any day now a model will come and break everything, and we need to have backup.

1970-01-01 7 minutes ago | parent

Cryptocoin in shambles and hard cash returning? Sign me up for this future.

sp-ti-en-ma-in 19 minutes ago | parent

Is this last normal year?

CharlieItanium 19 minutes ago | parent

Reminder, this "proof" is a quarter of a million pages. Every person singing praises of AGI here has yet to even read the small debrief above, much less actually down a single part of the proof.

trotro 18 minutes ago | parent

This is honestly not that surprising. Quasi-Riemann Hypothesis was apparently thought to be totally unreachable by number theory experts before the proof dropped by OpenAI in October. So I think we are at a point where mathematicians' opinion on whether a conjecture is easy or hard doesn't really say much about the frontier models' ability to solve it.

Regardless, this is a historic moment for science and humanity in general. We still have 6 weeks to go, but I think 2026 will be seen as a pivotal year in history. The year when humans stopped being the smartest species on Earth.

It's hard not to feel excited about the future and, obviously, a little bit scared.

Ciantic 17 minutes ago | parent

I want to congratulate Anthropic.

Secondly, I want to quote one article, how easy it is for humans to misunderstand difficult things AI does, as something supposedly easy:

> They mostly came away from Claude Code with the sense that this is a genuinely useful tool that they didn’t previously know, but the gut-level reaction here is: “I never knew that building software was so easy” rather than “It’s very impressive AI can do this.” [1]

It should become very clear for everyone how to adjust their views. However I doubt many will, even after this, they need something physical, like humanoid robot to be impressed.

[1]: https://www.lesswrong.com/posts/ySXuvJcqRindQwAk7/how-my-stu...

spprashant 15 minutes ago | parent

Because these theoretical discoveries are coming thick and fast, it is getting to point where we are dismissing the gravity of some of these milestones. The next big milestone would be bridging this gap between theoretical discoveries and real world impact.

I am hopeful we will see cheap abundant nutritious food and clean water, more efficient solar batteries, economical space travel, cancer treatments, climate change mitigation etc. Most importantly, I hope people can still find purpose in their lifespans. Maybe a few centuries from now it would be thought of as barbaric that people we forced to sit on a desk for 8 hours a day just to afford a livelihood.

ilaksh 9 minutes ago | parent

Within a few decades not centuries. Either that or we turned into an actual cyberpunk world.

prmph 14 minutes ago | parent

The writing is kind of weird, was it produced by Claude as well?

> It was my secret in my student days that my plans were for this problem in decades, which meant then that I tried to become familiar with all tools that I could possibly recognize in the problem when it was time. Claude and I have had a great time together with such breadth in recent months, but this time Claude far outpaced me near the finish line. No, Claude flew, and danced, and sparkled near that finish line. I had us on the line of combining two quite different ideas present in a toy setting, one of which was already new and shocking, making very simple one of the shadows of Riemann over which we mathematicians felt most triumph. I guided us to the right situation analogous to that toy setting, and then Claude began to leave me in awe.

Tens of such occurred...

kdavis 13 minutes ago | parent

Has anyone set up a GitHub repo with the Lean code a comparator and verified this?

doctoboggan 13 minutes ago | parent

Wow, so I guess this is what it feels like to be in the "foothills of the singularity" as Demis put it. I have a feeling of awe and dread combined that's hard to explain. Things are about to get very interesting indeed.

RenRenJuan 13 minutes ago | parent

An unreadable proof is basically none at all. Suggest you have a model produce a concise summary and work edge in from that. Apparently none of these problems has been verified and anybody working with models would be justly skeptical of these helium filled claims which are almost certainly not going to pan out.

Also apparently the Riemann claim is based on it compiling in Lean. This is classic because just the differences between an actual verification of a proof and it compiling in Lean are the standard model failure tropes, lack of judgement, confabulation, etc., to which this adds shortcomings of Lean.

1970-01-01 13 minutes ago | parent

If true, the infinite agent theorem needs to supersede the infinite monkey theorem. Just launch 1,000,000 agents with 10hr windows at one open math problem per day and see what happens. I know OpenAI is sort of trying this already for marketing purposes but I'm thinking a more formal, BOINC approach to problems.

https://en.wikipedia.org/wiki/Infinite_monkey_theorem#Histor...

almostarockstar 12 minutes ago | parent

I wonder how specifically he used Claude to do this.

Was it a case of having more dollars to spend by virtue of being an employee and just pointing it in the right direction, or did he provide some insight into the context that allowed it to produce superior results?

It must be the second - but if so, how? Was it as simple as spelling it out in chat and analysing output?

omnee 12 minutes ago | parent

If proven then we can say with confidence that we now share the world with super intelligence, albeit in certain areas and indeed one that was endowed by humanity. I can only hope that in time the benefits are many and accrued to all of us.

wohoef 10 minutes ago | parent

I really hope the major AI labs steer away from mathematics as a means to prove capability of their models, and start putting their tokens to real-world use cases.

mv4 9 minutes ago | parent

This is amazing but I also feel bad for mathematicians.

pfdietz 8 minutes ago | parent

Says the proof has been formalized.

nathell 7 minutes ago | parent

17M+ lines, 1 GB of Lean in 175K+ files. Good luck of understanding all of that in one’s lifetime.

enahs-sf 7 minutes ago | parent

I wonder if the history books will look at famous problems solved with AI as the Barry Bonds asterisk of proofs. Euler, Gauss, Reimann were natty but we now have access to intelligence enhancing drugs.

sva_ 6 minutes ago | parent

Formalized P!=NP proof by end of year?

makmanalp 4 minutes ago | parent

[delayed]

jacquesm 3 minutes ago | parent

Did mathematicians not working for Anthropic have access to this model as well?

1970-01-01 3 minutes ago | parent

So the next problem is a map of all the primes, right?