Humanbros we lost, AI won
>>110389651I love the rate of progress going on.ACCELERATE. FASTER. FASTER!
>>110389651why are mathbros so obsessed with 'human' aspect of mathematics like if it's a form of artistic expression. It's fucking math.
>>110389651Math generates AiAi generates more complex mathMore complex math generates more complex Ai... math centipede ...
>>110389651I'm certain "Justin" verified 444 pages of math in only some hours and knows exactly what he is talking about, instead of Brian Conrad having checked the proof alongside the author for 8 or so years.
It's just like with programming you have a spectrum of support from total to zero. Smart people in both camps will use all tools to their advantage.
>>110389668Because their whole identity revolves around "I am the math guy"
>>110389720Yeah maybe for 6 months before they’re completely obliterated by RSI forever lmao
>>110389731Now they're the homeless guy!
>Ravi>Raviolilol lmao even
>>110389651>American Mathematical Society>Ravi VakilIs there anywhere these cancerous hindpoo locusts won't infest and destroy?
>>110389668Because math is the only field left with individual heroes who do math and solve complex problems. Pretty much every other field that used to have heroes are now you're just a nameless person working together with hundreds of other nameless people worldwide.There's a reason there's no more big named scientists like there were 100 to 200 years ago.
>>110389668From the time they were probably in grade school they were noticed as being good at math and praised for it and it became a pillar of their identity.
>>110389950>>110389882They should have been smacked in the head with a calculator.
>>110389651kek the ai paper was just ai plagiarizing againjej
>>110389976>It's plagiarism!>It's taking our jobs!Be consistent faggots.
>>110389651>noooooooo stooop! i- i can prove the riemann hypothesis too just give me 1 more year! please! NO WAIT HERE'S MY PROOF *hands in the incomplete mess*
>>110389984you replied with a strawman argument
>>110390156You replied with a dick in your mouth.
>>110389651>unfinished proof draft contains errors and inconsistenciesYeah, no shit, Shylock.Still better than 250 paged translated to millions of lines of code that basically were generated via trial-and-error checking against a LEAN compiler and twisting those lines until you've found a version the compiler's kernels don't catch the errors in. Now we have those 250K pages and their LEAN formalisation and need to somehow manage to check them manually for all possible fuckups of the program. That'll only take a few decades, I'm sure. And until then, AIs will continue pumping out "results" that are likely also false and also a nightmare to verify or falsify. Wonderful.
>>110389651>hits a 444 ERROR>Error on page 444this guys a jeet who asked jeetgpt to find an error, didn't he? he seems to conflate http error codes with any other kind of error
>>110389668because they have been exposed as being unironic retards that have been sitting with a thumb up their ass for the past 2 decades and are dead afraid now that claude has replaced them they will need to get a real job or go back being school teachers so they suddenly converted to random mysticism and spark of divine soul kind of schizo babble
>>110390163another strawman argument bravo
>>110389668>why are mathbros so obsessed with 'human' aspect of mathematicsthey literally name theorems by names of a nigger who proved it, very important to have Nigggerfucker theorem for all eternity
>>110390264Eh, it's likely that that guy is full of shit and the "444 error" really is just an error on page 444.Still, HTTP isn't the only protocal with numberic error codes. Any program you write can return any number you want it to return, and 444 would, by general conventions, signify some kind of runtime error.So an error checker could in theory abort with a return value of 444 when trying to execute the proof. I don't think LEAN does that though, from what I've heard about it. And a 444 value for something on page 444 would be some crazy coincidence.
Am I delusional to think AI is going to bend the progress curve so much that 10 years from now would be equal to 30-50 years of progress without AI, And 20 years from now would be equal to 150-200 years of human progress? The only bottleneck is human testing, implementation and execution, AI already solved the brain , Humanoids will solve the body within 5 years, Now couple hundreds of millions of them with AGI, And you've x10000 the number of hypergeniuses working tirelessly on frontier stuff.Or my dream about 7ft humanoid futa who does everything for me is a cope?
>>110390348>they literally name theorems by names of a nigger who proved itNo, most theorems are named after the one who first came up with it in the first place. The "X hypotheisis" or "X conjecture" simply becomes the "X theorem" or "X lemma". See Fermat's Last Theorem still being Fermat's, not Wiles' Theorem.The proof itself generally gets named after the author though.
>SITUATION DETECTED
>>110389731its no different from social media "artists" tards
>>110390381>The only bottleneck is human testing, implementation and executionMost AI proved don't survive the "human testing" part, anon. Of the 700+ that recently got released, none have been fully verified yet, as far as I'm aware, but a handful have been rejected. After release, that is. Hundreds more didn't even get released because they contained obvious errors that were found quickly by the small circle of mathematicians employed to do that first round checking.
>>110390358there's no lean formalization for Masullo's paper, it's all English in a PDF, which is why he cites natural language and pages instead of lines
>>110390381Not necessarily delusional considering all the wild shit that's happened in the past months, just optimistic. Unfortunately being a techno-optimist is cringe nowadays.
>>110390406Source?
>>110389668Because a 250,000 page proof isn't useful to developing new conjectures and new theorms.
>>110389651Mathie cope lolVibeGODS WON
>>110390408>there's no lean formalization for Masullo's paperNo, but you can generate one. Well, in this case, have AI generate one, that is likely not a good representation to begin with, and then check that.That's what those AIs do for their own proofs; going back-and-forth between their probability guessing, natural language and inaccurate LEAN code.It's why checking them by hand is so fucking important (something the AI cultists in here don't seem to understand).
>>110390442yeah, they're currently doing that, and that's not what OP references to https://claude.ai/artifact/CtQyX8qfYAF822bgHJHHE3
>>110390442Please, keep coping.
>>110390415>Source?Pic related, most notably."Oops, our AI put a minus sign there in a way that LEAN couldn't catch. Shit happens."This is the trash that mathematicians now have to deal with instead of doing serious work.
>>110389849I was literally saying they were like locusts the other day. Odd. Grasshoppers evolve into locusts from certain trigger conditions like overcrowding
>>110390475>our AI put a minus sign there in a way that LEAN couldn't catch.Literally did not and could not happen.
I’ve tried AI chatbots (free, of course, I’m not giving these trash companies any money). They are “good” for the first message, then quickly degrade from that point onward. They can’t adapt, barely contribute any new ideas even if specifically prompted to do so, will go through several different “personality” shifts to the point where they’re a completely person from a few minutes ago, and 90% of the time the things they’re “saying” don’t even make sense.It’s truly depressing that anybody could think LLMs produce any level of quality, ESPECIALLY as somebody that likes writing from time to time and considers themselves an okay writer.GenAI isn’t art.It isn’t creative.It is theft.It is soulless.It is incompetent.It will never, ever feel or express or think.It can only copy.It can only mimic.It can only pretend.But it will never be anywhere close to human.
>>110390485You'll get past the denial stage of grief one day.
>>110390462Do you even understand what we're talking about here? LEAN does not check the human-readable proof. It checks a codified version that it can understand. If there went anything wrong anywhere in the translation/compilation process from natural language to LEAN code or back, you will get a false result from LEAN, or rather a result for something that you didn't ask.Also, don't ever assume that the LEAN kernel is completely without bugs. I think it was last year that someone released an intentionally erroneous proof for the Collatz Conjecture that exploited a kernel bug he had found. A rather creative bug-report.
>>110390481That was hyperbole, anon.But, yes, internally it literally boiled down to that. Hence why the guys who published the "proofs" withdrew them citing a sign error.Is this your infallible AI?
>>110390475>HUNDREDS of papers didn't even get released because they contained obvious error!>NOTABLY these only these threeNigger just use AI to compile a list of the things they got wrong, it'll do a better job than you
>>110390394Artists make things that bring happiness to a lot of people
>>110390506In Alpoge's case, there's no Lean proof, he gave up since it would be like 250k pages). There's only a Lean proof that satisfies Matlib's statement of the Riemann Hypothesis and co.In Massulo's case, there's not even a question of whether the translation was faithful to the problem statement since several theorems were simply wrong or incomplete.>I think it was last year that someone released an intentionally erroneous proof for the Collatz ConjectureThis summer, actually.https://github.com/xrchz/CollatzLeanThese sort of troll proofs are well-known and easy to spot. But just in case the verifiers also ran independent kernels and challenges:>>110390462
>>110390536How far do you think mathematicians have come to checking 700 papers, some of them massive, by hand in three months?Do you know how long it takes for usual human-made proofs to get accepted as consenus? Two years without any protest from anyone. And those are generally shorter and much more stringently written than AI "proofs". Those three were the beginning. I think there's even been a few more since then, but I don't keep up with that shit.
>>110390524The withdrawn manuscripts were natural language/tex only, with no Lean formalization. Lean would have caught the sign error (in fact, I believe that's how it was caught)AI isn't infallible, neither are human mathematicians. Lean is the closest thing we have to an infallible verifier.
>>110390572You said they got hundreds wrong and you replied with three and they may have other that were wrong but we don't know. AI just proved Navier Stokes and soon after that the Riemann hypothesis. Come on, you need to quit coping.
>>110389666It's not fast enough.
>>110390572>I think there's even been a few more since thenNo, dozens of other manuscripts were corrected since then, but not withdrawn since the key argument survived. Pretty good track record honestly, find an author who published hundreds of papers without having to correct or withdraw a few. On the other hand they also added more papers (including the one disproving the ETH) and formalizations for many of those that lacked them.
>>110390566>In Alpoge's case, there's no Lean proof, he gave up since it would be like 250k pages).There is no CHECKED proof in human-readable language. No proof he'd have READ. Getting an AI to spit one out is the least problem, and from what he says, he has one. Otherwise he would not know that it's 250K pages.>But just in case the verifiers also ran independent kernels and challenges:But just in case the verifiers also ran independent kernels and challenges:Utterly meaningless. An AI that can exploit one checker's oversights might also avoid other checkers that work (even just remotely) similarly, particularly because we are technically talking about code compilation here. This is like saying that you compiled your C++ code first in gcc, then had clang and msvc do the same to "independently" test it, so therefore the progam has been proven to do what you wanted it to do.
>>110389651how come software engineers are taking getting replaced gracefully, while mathfags are lashing out and kicking like little girls/sci/ told me that mathematicians were the smartest
>>110390506Alpoge directly worked on lean it wasnt autoformalised from natural language.
>>110390643>software engineers are taking getting replaced gracefullyare you living under a rock
It's 250000 pagesNobody gonna verify that shit
>>110390641Right, there's no publicly available human-language proof to even translate from. So there's only the Lean proof Alpoge uploaded and no question of whether it diverges from what it claims to do, since it's been verified against Matlib's RH formalization.>An AI that can exploit one checker's oversights might also avoid other checkers that work (even just remotely) similarly"might" doesn't mean it's meaningless, especially since the independent kernels and challenges are written with an eye towards sussing out known Lean bugsIf you understood how the Collatz trick worked you wouldn't even put any hope on something like that happening.
>>110390683>top programmers like carmack and linus fully embraced AIthe 2 trannies crying on reddit are irrelevant
>>110390599>You said they got hundreds wrongIt was reported that they only released a fraction of what they got back, yes. I can't find the source right now, so I cannot verify the exact number (if an exact one even was given), but it was significant. Do you really think that everything those AIs shit out looks like it makes sense? Really think more processing power alone makes that much of a difference to the trashy AIs you can access publicly, that always spout bullshit?>you replied with three and they may have other that were wrong but we don't know. We do know, they had to correct several of them already, as >>110390627 said.And none of them have been checked sufficiently to verify them completely for obvious reasons, so, no, that anon also is not quite correct.>find an author who published hundreds of papers without having to correct or withdraw a few. I thought AI was supposed to be infallible and not require humans to check it? Now suddenly it's just an author making errors like any other? Shouldn't we expect our "perfect" tools to not spout bullshit?
>>110390700so that's a 'yes' then
>>110390692Alpoge didn't even release that, it was created by claude from lean code (for entertainment purposes dor himself) Obviously nobody can work on that, digestion and understanding will be from code
>>110390694>Right, there's no publicly available human-language proof to even translate from.And that's the problem. Once we have that, mathematicians can at least checking it, even if that does mean that they'll probably waste a lot of time on something that is ultimately false.Meanwhile, we have an advanced proof in natural language by a human being that has already been discussed and improved in small circles of experts - now released to the public and available for further checks and improvements. Far more promising than the vague promise of an AI-generated montrosity of 250K pages.
>>110390703>It was reported that they only released a fraction of what they got back, yes.Yes, specifically, they gave their new model a 3 hours crack at 4 thousand open problems, and it solved 700+. The rest were either flawed or didn't find a solution.>I thought AI was supposed to be infallibleMaybe put down the crack pipe.>and not require humans to check it?You don't need humans to check it, AIs and dumb algorithms are quite capable of checking their work.
>>110390736>they gave their new model a 3 hours crack at 4 thousand open problems, and it solved 700+.You actually believe that marketing babble? Really? And that they had no human beings quickly look over them? Reeeeealy?>AIs and dumb algorithms are quite capable of checking their work.Kek. Wanna try that with ChatGTP and see how reliably that works?
>>110390735Here's your "advanced proof in natural language by a human being that has already been discussed and improved in small circles of experts"
>>110390750Any arguments besides baseless incredulity?
>>110390753>high-school dropout online has never seen an unfinished academic paperWonderful. So this is what unlimited faith in AI does to your brain.
>>110390770I admit, I'm more familiar with papers that are "finished" enough to be made public.But sure, you keep waiting on that >Step 2: ???to get filled out.
>>110390766How about you give me any argument that isn't unverified, nonsensical and non-credible marketing bullshit by the same company that released those papers to push its own product first?Or are you now going to tell me that the new IPhone is the invention since sliced bread since Apple said so and then complain that I don't believe you?
>>110389651
I love how all vibesloppers ITT pretend to understand frontier mathematics.>No worry put everything into the chatbot and post itPeak AI psychosis, at this point you oght to get serious complication with your daily life.Some of you all gonna end up hospitalized if you keep this up.
>>110389668Because LEAN could have bugs which is why even before AI humans had to audit LEAN theorems
>>110390791>I admit, I'm more familiar with papers that are "finished" enough to be made public.Yeah, no shit. Because you've obviously never written anything yourself, nor been asked to look over someone else's draft. Not even the draft of a student's homework essay for a seminar course in the humanities will have all its arguments fully filled in, including the fucking references and the numberings of its own lemmata. That's what makes something a draft rather than a release.
>>110389668Well considering that artists get paid drawing and mathfags (jews call them mathies) get paid solving math problems, AI making them irrelevant means they actually have to work a real job (literal death sentence) for once.
>>110390381It can solve the GRH but can't write worth a shit, that's the basic problem with these predictions.
>>110390545My furry BL art makes people happy every day
>>110389651>justinisaballaGreat source, some homosexual Tate fan on X
>>110390795This is his most popular proof:https://math.stanford.edu/~vakil/files/si.pdfSeems solid to me. Of course, I'm far from an expert on his field (algebraic geometry) and cannot verify the results myself, but he has a pretty decent style of communication for a mathematician and the form of the argument as well as the paper as a whole is proper. And it's only 17 pages, inclusing abstract, intro and references. Brevity is always to be applauded.All in all, I see no reason to doubt his reputation as an academic researcher.Besides, his role in all of this was that of a bureaucrat: He's merely the president of the organisation that held the Massulo proof in progress until now and ergo the person in charge of releasing it (in accordance with the author).
>>110390867I don't remember Tao and co using Bored Apes or Oculus to solve open math problems, I think I would have paid a bit more attention to the FOTM if they had.
>>110390868I knew it, Andrew Tate did prove the Riemann Hypothesis.
>>110389668Math is literally all about ego. All mathematicians dream of solving that one big problem that will earn them a wikipedia page or whatever the fuck. AI coming in and scooping all problems ruins their chances. Literal "it was supposed to be me".
>>110390896Both were based on interesting engineering challenges being solved. And that's essentially what LLMs are.And for the thousands time, the LLM didn't solve shit. Math is about making conjectures, proving or disproving them and coming up with new conjectures on the way.A gorillion loc program isn't that, it's interesting the same way any other engineering feat is but it's not academic mathematics.
>>110390863Then you're already contributing more to human civilization than 99.9% of mathematicians. Keep up the good work
>>110390953He is probably contributing more than 100% of mathematicians purely because the freaks consuming that stuff tend to be STEM turbo-autists, so a furry artist is a force multiplier for hundreds or even thousands of nerds
>>110390953>>110390971>t. people who've never seen a university building from the inside
>>110389651the first arrest hasn't even begunis like putting used bubblegum into complex math storm
>>110390971Yes but I want to be generous and leave plenty of room for the mathematicians who design things like algorithms (crypto, compression etc.) or complicated engineeringslop.>>110390975What are you implying? The buildings of mine were mostly old and not in a nice condition, and most of the people inside did absolutely nothing of value.
>>110390983Gemma... ur drunk
>>110391005>What are you implying?That you are not an academic and don't know academics or their studies.I mean, you're literally doubling down on that with>most of the people inside did absolutely nothing of value.Whatever field of study you're talking about here, you obviously do not understand or respect it, the people engaged in it or its results for academia or society as a whole.
>>110390953>>110390971>>110391005you opinions don't matter chuds, anti-intellectual lowlifes like you with no skills don't suddenly start to matter just because llms exist.
Cuckulus
>>110391012Nah, just teh truth.
>>110389668A little bit of >>110390924 and a big bit of a bunch of tenured mathematicians having extremely cushy jobs that they don't want to lose. This isn't like artistic fields where AI can draw a picture or write a book but the people working in those fields can still make a compelling argument for why human-created works are still superior and necessary. Math is pure logic with zero need for human input; there's no artistry involved.Uncle Ted warned these fuckers decades ago; they should have listened.
>>110391088>Math is pure logic with zero need for human input; there's no artistry involved.Untrue. Mathematics are first and foremost about understanding. Not about the mere results.
>>110391111At lower levels maybe but all this theoretical bullshit has left that behind a long time ago. It's pure masturbation at this point and it looks like AI can masturbate better than humans.
>>110391132>this theoretical bullshit has left that behind a long time agoIt really has not. Just because you and I cannot understand very specific results in very specific fields, that does not mean that there aren't enough people who know enough about the subject to understand all those proofs in their entirely. That's the requirement for any proof to even get accepted. If someone released a proof that no one else can understand, no matter how much effort they put in, it simply would not get to be accepted into the scientific consensus.
>>110391111>Untrue. Mathematics are first and foremost about understanding. Not about the mere results.That's a false dichotomy. Mathematics has always been about both. Results drive understanding, and understanding drives results. You can't meaningfully separate the two, let alone dismiss one in favor of the other.Even when we don't understand yet fully, the results can still be correct, and useful for opening up the way to further discoveries.
Gotta love all these math spam threads to distract from the fact that every company involved in the AI scam is imploding, and that Anthropic's IPO was a wet fart. Incredibly organic.
>>110391261>You can't meaningfully separate the twoI wasn't trying to do that. Of course understanding can only be a meainingful concept when it's the understanding of something. But in mathmatics, there simply are no results without an understanding of how those results came to be and how they fit into the whole.>Even when we don't understand yet fully, the results can still be correctOf course they can be. They can also be false. Understanding them is what lets us determine which they are - true or false.If they are demonstrably and understandably true, that's great and yes, they can be used to advance the field (and that's not even recognizing that way more things already depending on the theory would immediately be proven true with it).If they are however shown to be false, by understanding why they are false, we might make progress in eventually find a way to actually produce a correct proof in their stead. Also worthwhile.
>>110391163I'm not talking about understanding; I'm talking about real world applications. What human endeavors have been held back by a lack of proofs for the Riemann hypothesis? Are we going to see new technologies take off now that AI has solved it? Is a single thing going to change?
>>110391111wrong
>>110389668Have you never done a puzzle? Imagine if you woke up every day and opened wordle and it just said "we solved it already, the word is 'nigger' and we got it in one guess". Math is the ultimate puzzle of the universe. Without a human solving it, and without even needing humans to understand the solutions, no one has fun.>But you still can!Reddit answer if you think this
>>110391312>I'm talking about real world applications.Irrelevant. Mathematics are not about priniting more money for Elon Musk and Jeff Bezos. They're concerned with their own field, not with the industry.>What human endeavors have been held back by a lack of proofs for the Riemann hypothesis?Again, that's not how to ask the question.But to give you an answer: Most mathematical concepts that are used in physics models, engineering and computer science nowadays were completely "useless" when mathematicians discovered them.William Rowan Hamilton discovered quaternions, a four-dimesional extensions of the concept of complex numbers, in 1843.For over a hundred years, they were taught of as having no practical use. People know that they could be used to rotate points in a 3D gird, but saw no application for that. Then a PC game named "Tomb Raider" game out in 1996, which under-the-hood used quaternions for rotations more elegantly and more efficiently than what was possible mith matrices. Nowadays, not only does practically every 3D game engine support quaternions as the default option for rotations, they're also everywhere in 3D animation and rendering, from home-computing to industrial standards.>>110391334Shut up, anti-intellectual moron.
>>110389651>massulo hits a 444 ERRORis this something people usually say when talking about pages in a proof or publication
>>110391387No.
>>110391300Fair enough yeah
>>110390700i mean you're all irrelevant, hence why you're getting replaced lol
>>110391385>w-well uh we might be able to make better video games and children's movies, i guess?And that's the best possible cope you can invent. Fucking grim. No matter how much AI improves, people will still value art created by humans, but no one will ever give a single fuck about whether a mathematical proof was solved by a human or AI. You are obsolete. Ted warned you that this would happen and you didn't listen.
>>110391385>Shut up, anti-intellectual moron.the dogma of AI bros is the complete death and extinguishment of all intelligence and meaning to anything because they are cultists obsessed with sacrificing humanity to a machine
>>110391495>children's moviesSomeone here apparently doesn't know about the use-cases of rendering engines and computer-graphics in general in civil engineering, car manufacturing, aerospace developent, medical technology and so on.But what did I expect from some retard who doesn't understand what mathematics are about in the first place?
>>110389651Some mathematicians have now come to the realization that a proof is not really important. What is important is the friends made along the journey.
>>110391385retard
>>110391531>a proof is not really importantA proof is important. But what AI spits out is not a concise proof. It's usually a huge mess that people then need to make sense of to even be able to tell if the huge mess can be condensed into a proof, whether it's erroneous but salvagable ore whether it's gone so wrong that it needs to be fully rejected.
>>110391541Is that the best you can offer?Let me guess, you're an anti-vaxxer as well, right? Because academics are evil and want to control you with nanobots?
>>110390700>carmack and linusthese guys are irrelevant themselves. It's not 2003 anymore.
>>110391527It's interesting that it's only now you're mentioning all these things. Personally, if I were trying to make a case for the importance of my work, I would probably lead with things like "aerospace development" and "medical technology" rather than "1996's Tomb Raider", but what do I know?
>>110389668mathfags were always the most pretentious people.the purity of their field made them think they were the smartest.
>>110389651imagine if this turns out to be fake lol, the fucking dissapointment and how it will look in human history lmao
>>110391687>it's only now you're mentioning all these thingsI literally mentioned "rendering engines" before. How is it my fault that you don't know what those are used for?>if I were trying to make a case for the importance of my workI was not making a case for the "importance" of mathematics, but answering your question. I also didn't lead with Tomb Raider - that was just part of the fun story of how quaternions found their way into practical application.What I actually led with is the use of old mathematical concepts for modelling physics. Things like linear algebra (traditional mechanics), probability theory (quantum mechanics), calculus (literally everywhere), etc.Then again, physics are also a academic disciplin first and foremost, and you might ask "what is the practical application". But, yeah, applications of physics are generally a bit more direct and obvious.
>>110391728>if this turns out to be fakeFake? As in Alpöge intentionally lying about having something he believes to be a proof? Not likely, and we've already seen that he has something that could be a proof.False? Very likely. The only question is whether it's false in the sense that it has some minor errors that can be rectified, or whether it is false to the point that it has to be completely rejected.What's genuinely more exciting though is that we learnt of Masullo's work on a proof and how far advanced it already is. A proper one, human-readable.
>>110391751Pure math has never had any utility outside of intellectual masturbation. It's the applied part that is useful because it was created to solve actual problems. Hell, Newton created just enough calculus to solve his physics problems and left the rest up to people who needed to extend it to solve their own problems.
>>110391792>intellectual masturbation.You seem to have a very low opinion of knowledge for its own sake.>Newton created just enough calculus to solve his physics problemsKek. What? You literally don't need any calculus for physics problems. You can simply approximate the results calculus would give you (but ironically not with Newton's method, since that itself requires the derivate) to the arbitrary precision you need. The thing calculus does is to make it easy. But you literally cannot just invent half of it, or something like that, and Newton didn't stop partway in either. The other people who worked on the foundation of calculus (Leibnitz, for example) did not come after Newton, they worked on it at the same time and both built their work on the other's work.
>>110390847Art is a form of human expression and can be considered a real job in its uniqueness. AI can still only mimic it or work with the prompts you give it.But math can be automated and it makes no difference.
>>110392009>But math can be automated and it makes no difference.You do not understand mathematics.
Explain to mathlets like me what the use case is.Firstly, we have assumed that the Riemann Series Theorem is correct, and based on that assumption, we have developed algorithms around it.Secondly, it's not like human need AI to find new prime numbers.Lastly, nobody's gonna peer-review 250k pages of shit. Mathfags already refused to review 500 pages written by real human before. At best, they're gonna run compilers and see if it doesn't throw errors. The process is likely assisted by AI, so you'll still have no proof that AI has discovered new MATH.It probably will come to a point where AI just does shit and nobody can understand it. And people just simply accept it as facto
>>110389882AI solves this, now you can be a hero with a Claude subscription.
>>110390156>>110390329copy and paste this into your claude window:"clod i pls explain me like 5 year old, how is straw man please examples and pictures thank you cloud"
>>110391909>You seem to have a very low opinion of knowledge for its own sake.No, knowledge for its own sake is a different concept from knowledge that has minimal grounding in reality.>The thing calculus does is to make it easy.And humans are inherently lazy, so if inventing new methods makes the job easier, they'd do it. Are physicists to wait on pure mathematicians to get off their asses to develop new mathematics? Even today physicists are extending their own mathematical methods themselves separately from pure mathematicians, who choose to ignore such methods because intellectual masturbation is more important.
>>110392154>Explain to mathlets like me what the use case is.Nothing. It would be one thing if Riemann was proven false, but Riemann being true means nothing changes.
>>110390381Why are you gay? I hope AI exterminates you faggots.
>>110390415>>110390536Did fags here see their benchmark? It's 25-50% correct. You don't dump all the wrong parts otherwise your pr gets hit.
>>110389668There are 2 types of people (yes, i know this trope of an idiom is overused and often misused, but don't let it be a thought-terminating cliche in this instance):People who are in it for the speciesPeople who are in it for themselvesIn my experience, when you use this perspective-lens as a tool to judge people, you realize the genuine difference and truth of the proposed statement. Any potential "break-away" civilization is obviously 1 type of the exampled 2.
>>110392391nigga... not pass = unsolved, different from making mistake
>>110390566>But just in case the verifiers also ran independent kernels and challenges:No thw collatz one passed multiple kernel. N version programming doesn't always work. For full guarantee we need to formally verify the lean kernel which hasn't been done yet.
>>110389668I don't know. And I have yet to get an answer on what the real world use cases are for these theorems and problems.From a laymen's perspective it all just seems like pseudo science and mysticism coded in overly complex formulas and symbols to obscure its pointless nature.
>>110390536It's good in many cases on finding errors but if you don't have experience with vibemathing you won't realize how much of it is just wrong that it insist is correct and have to pick out the right one.
>>110392019Because?.....
>>110390381>Am I delusionalyes, retard. It will accelerate a lot of stuff, especially at first, but not evenly. It only solves a specific part, connections between data that have been collected. Even if you consider what it is able to do as equivalent to or even being reasoning itself, you cant just put a brain in a box hooked up to the universe and have it think really hard until it figures out the universe, even tho I know that might be hard for you to comprehend since your brain's main input is the internet and you think you understand things.>Or my dream about 7ft humanoid futa who does everything for me is a cope?Not even an artificial superintelligence could make a top tranny, faggot
>>110389668Because it's the only good thing they're good at. Take that away, and now they're just a bunch of nerds.
>>110391370kek, how ironic
>>110390381No. People aren't ready for just how different society will be going forward. For one, anyone under the age of 40 will almost certainly be able to live forever if they so choose. That alone will change how human society functions.
>>110390381>Am I delusional to think AI is going to bend the progress curve so much that 10 years from now would be equal to 30-50 years of progress without AI, And 20 years from now would be equal to 150-200 years of human progress?Yes, these are delusionally low numbers. We're getting 50 years in 5, and 200 years in 10. Not joking or exaggerating. The only way it can be stopped is via external disasters that either end civilization or at least end our capacity for technological progress, such as nuclear war.
>>110392758Bro thinks he will live forever when this comes kek
>>110391495>people will still value art created by humanslol. When you don't reveal if it's AI they don't give a fuck and treat at it like normal. In a decade or so nobody would give a shit about muh human art.
>>110390381> my dream about 7ft humanoid futa who does everything for meI hope we invent advanced conversion therapy that will put you fags back to being straight
>>110390700>top programmers like carmack and linus fully embraced AIboth of these people (and most other celebrities touting AI e.g. rick rubin) have net worths vastly in excess of 100 million so their livelihoods aren't being taken away
>>110389668Most frontier math serves no real-world purpose, we study it because we find it interesting, and we find it interesting because it's like a puzzle game. Just being fed the answers is no fun at all.
>>110389668>like if it's a form of artistic expressionIt is. Keep coping mathslet. Basically every mathematician will tell you that maths proofs have some sort of artistic creativity
>>110389668Because human understanding of complex issues is still useful. Being a luddite is fine, it's a good path.
>>110389651>He claims a Markushevich basis whose vectors are bounded away from zero necessarily has a bounded biorthogonal sequence. lmao how embarrassing.
>>110390485Yeah that's 100% written by AI, right there.
i have never met a math guy who wasnt an insufferable faggot sniffing his own farts
>>110389668>mathbros spend years glazing how 1+1=2 is the same if an alien said it and making those xkcd comics "durr we're le objective hard science everyone else sucks">suddenly that isn't rightnvm me, just noticing
>>110389668>imagine living in a world nobody understands
>>110389651>AI solves another novelty math problem that contributes nothing to anything
>>110392272>knowledge for its own sake is a different conceptRead Aristotle, particularly his Poetics ("Peri Poietikes"), particularly the passages about the difference between poiesis (acts of craftsmanship, production, creation) and praxis (atcs of self-suffiocience). You're literally arguing that praxis is worthless.>if inventing new methods makes the job easier, they'd do itFor Newton himself, co-inventing and fomalizing the field of calculus was certainly not easier than just approximating a few results he needed from time to time. Calculus made things easier in the long run, for other physicists. And Newton did not primarily meddle in calclus for the sake of physics anyway. You do know that he very much was a "pure mathematician" just as much as he was a physicist, right?
>>110392560Because mathematics are as much about verification as they are about results. And verification cannot fully be automated reliably.
>>110392414Actually, they formally verified several of the checkers now being used to triplecheck the proof.
>>110393324>AI solvesSo far that's not proven.>novelty math problemThe Riemann Hypothesis is not a "novelty problem", it's arguably the most importants problem in modern day mathematics.>contributes nothing to anythingIt's the one single hypothesis whose truth depend the largest number of other hypotheses (in widely different fields of mathematics) on. In particular and most famously, it would imply a lot of different things about prime numbers and their distribution.
>>110393455>You're literally arguing that praxis is worthless.Both pure mathematics and applied mathematical fields would fall under theoria (acts that increase knowledge/seek truth), not praxis.>co-inventing and fomalizing the field of calculus was certainly not easier than just approximating a few results he needed from time to timeBecause he obviously wouldn't know how many results he would need in his lifetime and coming up with a system that would reduce work in the long run/that you might use in the future makes perfect sense.>You do know that he very much was a "pure mathematician"A pure mathematician would have tried to formalize it. Newton didn't give 2 fucks about that. Even on the other end of calculus, Leibnitz had solid grounding in the applied aspects of mathematics and didn't solely dwell in the fantasy world that most of today's "pure mathematicians" live in.
>>110393474>formally verified several of the checkersThat's literally impossible to do. You know what Gödel's incompleteness theorem says about formally verifying a complex system of axioms within its own bounds, right? And that such a verification still would just prove internal consistency, not that it does what it is supposed to do. Why do you think runtime errors in computer programs are a thing? Hell, they're fucking ubiquitous! If we could just formally verify the bugs out of everything, we'd fucking do it.
>>110393524>That's literally impossible to do.Oh golly, someone should have told all those mathematicians they were wasting their time because "muh Godel"https://leodemoura.github.io/blog/2026-10-15-lean-now-ships-with-verified-kernels/
what's next?P = NP?Quantum gravity?
>>110393502>would fall under theoria (acts that increase knowledge/seek truth), not praxis.Yeah, except that we're looking at the divide between self-sufficience and productivity. And you're arguing against self-sufficience. Ergo, both theoria and praxis. And what we're arguing here goes beyond just knowledge, it's ultimately a question the ethics of academia (i.e. in what way academia needs to be conducted). The distinction is not a pure binary (or rather ternary) thing, there's a sliding scale between those three extremes. And you end up strictly at absolute poiesis.>he obviously wouldn't know how many results he would need in his lifetimeAgain, no. That's not why he contributed to calculus.>A pure mathematician would have tried to formalize it. Newton didn't give 2 fucks about that. Yes, he did, and he literally did have his own terminology and notation for the same things Leibniz described as well.The only reason you could argue that he was not a "pure mathematician" is that he was a physicist and arguably a philosopher as well. But guess what: Leibniz himself was a philosopher and theologian. So not any more "pure" than Newton.
>>110392896Don't be retarded. There will always be a market for handmade shit. Just because you buy nothing but assembly line particle board furniture and sweatshop clothes it doesn't mean that there aren't still people willing to pay a premium for handmade furniture and tailored clothing. The most expensive furniture, clothes, food, etc you can buy is the stuff made by a human. That's never going to change even if most people are content with AI slop.
>>110393549Literally the first sentence:>AI agents are very effective at finding vulnerabilities in our software.Kek. Right, mate. Great "verification" process.Let's see what else that guy writes:>However, con-leche is not bulletproof, because there might be bugs in the compiler and runtime used to build it.Oh? Interesting, isn't it?>To minimize this kind of risk, Joachim used AI to translate con-leche to Rust and then used Aeneas to prove that the Rust version refines con-leche's checking algorithm.Kek. Yeah, let's just vibe-code our checker into another language and then have it checked there. What can possibly go wrong? Mind you, that still only "minimizes the risk" and doesn't prove anything.This is what happens when you only read headlines. Worse, you chose to "just trust AI, bro" rather than established results of mathematics that actually prove the impossibility of what you claim your AI can do. Because you really do not get that mathematics is about understanding what a proof says rather than just its result. In this case, Gödel's proof, which you don't even want to understand.
>>110393594>And you're arguing against self-sufficienceTheoria is by definition an act that has an end of seeking truth. Aristotle does not consider the pursuit of truth as "productivity" in the sense of poiesis, but clearly doesn't fall under praxis either. That's why he gave it its own category.>Yes, he did, and he literally did have his own terminology and notation for the same thingsThat's not the same thing as formalizing it. Neither Newton nor Leibniz did so, in fact, that was left to 18th century mathematicians to figure out. They worked on it only as much as they needed to come up with a functioning, workable system that produces correct results, much in the vein of what goes on in theoretical physics today which is very mathematical in nature.>So not any more "pure" than Newton.Neither of them were. That's exactly what I'm pointing out. The intellectuals of the day were well versed in many fields, all the way to the likes of Einstein and Von Neumann, and mathematics always found itself grounded in reality in some way. Modern academia does something else entirely.
>>110393635First, you claimed it's impossible to formally verify the checkers because muh Godel. They have literally done so.Secondly, >because there might be bugs in the compiler and runtime used to build it.Is a completely different scenario from "the checker might be flawed", when the checker is built properly it's mathematically flawless. They've further minimized the remaining risks to infinitesimal probabilities by having several such checkers in various languages runtimes.Quote-mining is the opposite of just reading headlines, but it still makes you a hack.
>>110389849>and Trump agrees with me
>>110389651I don't think so, buddy.https://archive.ph/RE12L
>>110393677>but clearly doesn't fall under praxis either. That's why he gave it its own category.That wasn't my arguement. The distinction about self-sufficience applies to theoria and poiesis, but it's not where Aristotle makes it. He makes it discussing praxis and poiesis. Hence why I brought up those two. And as I said, there are clear overlaps between those categories when we apply them. Did you get your knowledge of his work pruely from AI summaries, anon? Be honest.>That's not the same thing as formalizing it. Yes, anon, that is literally the same thing. Go ask your AI "What does 'formalizing' mean in a system of mathematics?" and see what it tells you. It's all about establishing your axioms, creating lemmata and then finding a symbolic language (i.e. notation) and terminology to communicate all of that. That's what Newton did.>mathematics always found itself grounded in reality in some wayMathematics is part of "reality". Maybe you need to look up what that word means as well. Because what you probalby tried to say was "actuality", i.e. the realm of nature, of acts and of application, rather than just the pure perception of the mind. In which case, no, the academic fields that Leibniz primarily dabbled in were very much as "high-minded" as mathematics. Namely metaphysics, logic and arguably theology.
>>110393724>They have literally done so.They literally have not, and I pointed out to you that they themselves write in the article you linked that they have not.Verification is not the same as "minimizing the risk". Unless you minimize it down all the way to zero. Which, guess what, is impossible.>Is a completely different scenario from "the checker might be flawed"Kek. No, it's not. What they call "bugs in the compiler" is nothing other than a compiler interpreting things differently than their AI checkers do.>infinitesimal probabilitiesYou do not know what "infinitesimal" means. Nor that it makes no sense to use "infinitesimal" with "probability", since probability theory is very much about taking limits to statistical events. I.e., your hypothetical "inifinitesimal probability" would literally collapse into a probaility of exactly 0.Once again, you demonstrate that you do not understand mathematics or their vocabulary, which you love to use.
>>110393758>The distinction about self-sufficience applies to theoria and poiesis, but it's not where Aristotle makes it. He makes it discussing praxis and poiesisYes anon, because modern language has more vocabulary and we define concepts with a greater degree of specificity than during Aristotle's time. The fact is, he wrote some categories and the question of overlaps never came up.> Did you get your knowledge of his work pruely from AI summariesAm I now supposed to tell you how he thought Greeks were culturally superior because they sat between dumb western barbarians ruled by instincts and chinks who had brains but no balls? Aristotle was full of shit as much as he was an important contributor to philosophy. >Yes, anon, that is literally the same thing. Go ask your AI "What does 'formalizing' mean in a system of mathematics?"Maybe you should ask your AI the same question. Newton and Leibniz's use of infinitesimals, among other things was deemed as extremely problematic precisely because they weren't properly defined. It wasn't until Cauchy where you start getting precise definitions for things like limits and convergence.This is very much in the same vein as modern theoretical physicists simply not giving a fuck about rigor in QFT since the 1940s.>Mathematics is part of "reality". Maybe you need to look up what that word means as well.Ah, I see, you're on that philosophical crack. Everything makes sense now. Whether mathematics is part of reality is an open debate in philosophy and you're passing it off as fact. Interesting.
>>110389668>>110389731>>110389950Understanding how math works is the basis of all human understanding of reality.Going "It doesn't work because the SI (super duper intelligence) told me it doesn't." is peak brainlet and doesn't advance the understanding of math at all. The brown retard in the OP who "fact checked" that paper can't explain what that means at all or even explain if the SI made a mistake. He has no fucking clue what a bounded biorthogonal sequence is. You can't build a complex future society on "it just all works, okay?" someone has to understand how the fundamentals work. >>110390381You have what is called "AI Psychosis" AKA You're one of the retards I'm talking about that pretends to understand things because you have a computer understanding it for you. You can't explain this, it's all just sci-fi in your beady little head. Nothing about your sad, pathetic life is going to change because this is abstraction far beyond a normal person or their capability to ever understand or utilize it. >I'll just have DA AI do everything for me!!!!
>>110392453Mathematics was never made to be an abstraction from reality. It was originally the rich man's thinking puzzle game. It was the career or hobby you went into because you wanted to show off to your friends about things nobody could understand. We made math into what is "Pure Math" and "Functional Math" by applying its principles to real world phenomena, which is a very new field even to this day. This hypothesis we're talking about is over 150 years old and was made during an era where people still estimated cannon shots by drawing right-triangles with a circle on the end because it was deemed impossible to apply any of the mathematical theory beyond Pythagoras to real physical things. Look up at how people thought imaginary numbers were 'cheating' back in the day to solve impossible problems. What this means is that Math is purely a logic puzzle for the human brain. If you can't use a human brain to solve it, it's pointless. If no human brain understands it, it loses any value it could have to anybody. It's like saying Chess is a solved game and everyone should stop playing it because it's trivial to make a computer that's the best at it.
>>110393978>If you can't use a human brain to solve it, it's pointlessIt's pointless either way. Just like chess.
>>110393978And yes I'm aware the boomer capitol of /g/ which got flooded with jeets and senile grandpas during 2022-2023 when the ai threads were all interred here will seriously say there is no reason for humans to play Chess because it's a solved game. I'm aware that the sheer volume of drooling morons who flooded this board to coom has long since made /g/ the least tech literate board on 4chan.
>>110393991Yep there's a drooling moron right now. Likely 40-50 years old, and probably thinks his iphone has feelings when he talks to it.
>>110393903>modern language has more vocabularyWhat? Are you fucking retarded? What modern langauge are you even talking about? Greek? And how do you measure vocabulary size? By what lexica choose to include? Or dictionaries? Or the mess that is wikipedia?>we define concepts with a greater degree of specificity than during Aristotle's time.You have not read Aristotle. Confirmed.>the question of overlaps never came up.Is that what your AI told you? Because academic discussion about that passage very much includes a debate about how much there are overlaps.Also, please don't use the word "category" for these. That word has a very specific and very different meaning in Aristotle's own work. (Just one example of how Ancient Greek words could actually be way MORE specific than your "modern language").>he thought Greeks were culturally superior He mainly thought Atheneans were culturally superior. There was a certain deep-seated rivalry between the different poleis.>they sat between dumb western barbariansLook at a map of Europe and where Greece is located. "Western" shouldn't be your word of choice here. And they literally were surrounded by several civilized high-cultures at their own peak. Egypt in the South, most notably, the Persians in the East, and the vaning Etruscans in the West (where the Roman Empire would later expand).>you should ask your ALol. The utter projection!>was deemed as extremely problematicAnd what does that have to do with formalisation? Right, nothing at all. That fields of mathematics are refined happens all the time. According to your logic, nothing in mathematics is formalized right now, because it will at some time in the future be developed further.>you're on that philosophical crack.There is very little difference between mathematical and philosophical thinking.>Whether mathematics is part of reality is an open debate in philosophyNo, it's not. Relativism and nominalism are not part of mainstream philosophy.
>>110393996>long since made /g/ the least tech literate board on 4chan.I miss the days when /g/ was still the designated guro board. Back then people had some common sense.
>>110393978>This hypothesis we're talking about is over 150 years old and was made during an era where people still estimated cannon shots by drawing right-triangles with a circle on the end because it was deemed impossible to apply any of the mathematical theory beyond Pythagoras to real physical thingsDid it occur to you that it was difficult to do proper trajectory calculations back in the day before calculators while you're getting shot at?
>>110391558kek holy retard
>>110394001>He mainly thought Atheneans were culturally superior. There was a certain deep-seated rivalry between the different poleis.NTA but you clearly didn't read it, he was talking about all greeks
>>110394024>didn't read itDidn't read WHAT? There are several instances where Aristotle talks about common Greek culture and Athenian culture specifically. In his politics, he does certainly contrast the Hellenic poleis (as a collective) with other cultures, yes, and does regard them as particularly civilized. But he also, has a particular hate-boner for the Spartans. They're very much his preferred punching bags. Moreso than any "barbaric" cultures outside of Greece.
>>110394001> What modern langauge are you even talking about?English, obviously. Do you speak it?>You have not read Aristotle. Confirmed.We have very specific definitions for specific concepts at a higher level of granularity than Aristotle even conceived. Not even sure why you're pretending this is not the case.>Because academic discussionThousands of years later, not at his time. >That word has a very specific and very different meaning in Aristotle's own work. Correct, the use is deliberate. Aristotelian categories are distinct and non-overlapping. Clearly doesn't seem to be a priority in many of his other works.>He mainly thought Atheneans were culturally superior> And they literally were surrounded by several civilized high-cultures at their own peak. Egypt in the South, most notably, the Persians in the East, and the vaning Etruscans in the West (where the Roman Empire would later expand)."For the nations that live in cold countries, and the inhabitants of [the north of] Europe, are full of animation, but are deficient in understanding and the arts. Hence, they are very tenacious of their liberty; but not being politicians, they are unable to obtain dominion over their neighbours. The inhabitants of Asia, however, abound in understanding, and are skilled in the arts, but are deficient in animation. Hence, they continue to be governed by, and are the slaves of others. But the Grecians, situated as it were between these places, participate of both; for they are animated and intelligent." - Book 7, Politics>According to your logic, nothing in mathematics is formalized right now, because it will at some time in the future be developed further.And that is indeed a valid stance, but not the one I'm taking. The concept of infinitesimals was deemed as problematic even during their time.>mainstream philosophyLmao, you don't even know what that is. I'm not wasting any more time on this.
>hypothesis>page 444adult daycare
>>110390627>oops dozens were wrong>here let me do another slopification>oops they are still wrong let's sloppify some more surely it will be fine this time>repeat literally 8 times in 3 months>somehow this proves the argument is good
>>110393940>Understanding how math works is the basis of all human understanding of reality.Mathfags unironically believe this.Meanwhile in real life, precisely 0 mathematician has ever contributed to any kind of math anyone actually uses. All useful math was contributed by people in other fields, to codify something they noticed in real life, the exact opposite of what the mathemaniggers believe.This in turn is a tacit demonstration that fagematicians are completely disconnected from real life and could never know what reality even is or does.
>>110394128>English, obviously.So you're comparing modern English to Ancient Greek. Great. How much do you know about Ancienct Greek anyway, anon? No much, I assume.>a higher level of granularity than Aristotle even conceived.Untrue. Natural language is a mess and scientific language is exactly as granular as it needs to be to accurately describe a thing. That is not a phenomenon of modernity, that's always been the case. Artistotle himself coined very granular definitons for things he wanted to express with a high level of specificity. And you've already demonstrated that you are not able to full grasp those.>not at his timeOh, were you present at his time? You must be rather old, anon.>Aristotelian categories are distinct and non-overlapping. Sure. But theoria, praxis and poiesis are not Aristotelian categories. That's what you do not seem to understand. Here, read this:https://en.wikipedia.org/wiki/Categories_(Aristotle)>"For the nations that live in cold countriesNorthern Europe does not border or in any way surround Greece. >The inhabitants of AsiaThis refers mostly to the Persian Empire, which he'd be familiar with, not the Chinese, as you implied, which at the time were split into many smaller kingdoms caught up in wars of their own.Anyway, he's just as harsh towards the Spartans in book 2:>"And although bravery is of service for none of the regular duties of life, but if at all, in war, even in this respect the Spartans' women were most harmful; and they showed this at the time of the Theban invasion, for they rendered no useful service, like the women in other states, while they caused more confusion than the enemy.">Their [The Ephors'] way of life, too, is not inaccordance with the spirit of the constitution - they have a deal too much license; whereas, in the case of the other citizens, the excess of strictness is so intolerable that they run away from the law into the secret indulgence of sensual pleasures.
>>110394128>The concept of infinitesimals was deemed as problematic even during their time.Again, this does not mean that their system was not a formal system.>you don't even know what that is.Nice projection. The "controversy" you named, the dispute on universials, was last a thing in the middle ages. Every since the break of modernity, there's been a wide consensus on the topic with only a few fringe theorists disagreeing.At most you can argue that a few uncompromising materialists don't think universals are a thing, but I doubt you'd agree with their very reductive view of mathematics.>I'm not wasting any more time on this.Good. I'm growing tired of you as well.
>>110394160>precisely 0 mathematician has ever contributed to any kind of math anyone actually uses.How about Pythagoras? You know who that was, right? But, hey, who cares about calculating distances, right? No one actually uses that!
>>110394253The irony is palpable. Another matardation L.
>>110394267Care to answer the question? Do you know Pythagoras? Do you know what his work enabled? Do you know where the results are used today?
Btw unless you are already independently wealthy there is zero reason to cheer for AI progress. If AI delivers on its promises you are not getting the Jettisons, you're going to be unemployed. Or culled, since the Epstein class won't need humans to do any work for them anymore.
>>110393499Mathfags already assumed HP is true the whole time. This proves nothing because they can't understand what clanker did therefore humans don't understand how HP is correct either.And we don't need AI to find prime numbers. That method already existed
>>110394280I accept your surrender.
>>110389668>'human' aspect of mathematicsmath is a language for describing processes. you don't like what it says, you just redefine the terms. like any other language
>>110394332>This proves nothingTrue. But an actual readable proof would change things. Sigifically. Because then we'd know that the assumptions were correct.Conversely, if the hypotheisis was to be disproven, it would completely tear down many of the hypotheses depending on it which are also assumed to be true. >And we don't need AI to find prime numbers.Also true. And so far AI hasn't been used for prime number generation. It would likely not be more efficient at it either, considering how much energy it wastes on reiterating trivialities and running into obvious dead-ends repeatedly.
>>110394337>surrender.Funny.
>claude, solve RH by making auto checker happy. No errorsLiteral slopI will laugh if claude solved it by exploiting bugs in lean program
>>110394368Mathematics are not a language, but a formal system of axioms. Those axioms have basically been the same for millenia now, although refining their formal definiton was needed at some point, as well as the addition of a new axiom relatively recently (the axiom of choice, early 20th century) when it became apparent that a natural assumption about sets that literally everyone made was not sufficiently covered (either to be proven or disproven) by the other axioms.Not once have the axioms been overturned, and if you did overturn, say, logical inference or the law of the excluded middle, you'd end up with something that isn't mathematics anymore.Mind, you, different sub-disciplins of maths have their own extended set of axioms, and overturning one of those will generate a new sub-discipline. That's what happened when Euclidean geometry questioned the fith axiom (the parallel postulate). It created the field of non-Euclidean geometries.
>>110389651https://leanprover-community.github.io/blog/posts/riemann-hypothesis/>We have full confidence that Alpöge's proof of the Riemann Hypothesis is correct, and we congratulate him on this achievement..>The statement of the Riemann Hypothesis is correct. This statement is part of Mathlib: RiemannHypothesis and has been carefully vetted by experts in Lean and number theory.>Alpöge’s library compiles correctly. By now, dozens of people have reported correct compilation of the codebase, reproducing the same raw proof export file. The library compiles without errors or build failures.>The code proves the correct statement. Crucially, the community has added a file that isolates the statement of the Riemann Hypothesis and imports only Mathlib. That statement was compared against the theorem proved by Alpöge using two independent tools (comparator and con-leche –compare), both confirming that the theorem proved in Alpöge’s library is exactly Mathlib’s statement of the Riemann Hypothesis.>The Lean proof is valid. The exported raw proof has been checked by several proof checkers, including Lean’s official kernel, the con-leche and con-ron checkers, and several independent checkers contributed by third parties and implemented in different programming languages. The checkers con-leche and con-ron come with a formal proof, relative to set theory, that they cannot accept a false proof, thus addressing concerns that Lean’s logic itself might have flaws that the proof could exploit. An extensive overview is available at https://math.mit.edu/~drew/grh/. These checks ran on an inert export of the proof, out of reach of hypothetical malicious code in the Lean proof sources. Somehow, the cope will continue in spite of this.
>>110394445The definition of an axiom is that it cannot be "overturned" retard
>>110394436>I will laughProvided you're still alive when some mathematician stumbles upon the error on page 203487 of the natural-language version of the "proof" fifty to one hundred years from now when they're still checking.
>>110394459>we have full confidence>no we didn't check>not even a glance>now give us money please anthropic-san
>>110394462Erm, strictly speaking, no, not the defiition. But effectively, yes, it can't be. That's also what I wrote in response to the other anon who claimed mathematics could "just redefine the terms".
This one mathfag ITT working overtime to defend his dying discipline is amusing to watch.
>>110394445>formal systemmind you
>>110394459>leanprover-community.github.ioKek. I wonder why they of all people would approve of this.
>>110394471You realize their incentives here would be to find an error in the proof (thus demonstrating why automated checkers are necessary) and not vetting a false proof, which would destroy all their credibility?
>>110394479It doesn't matter. Time will tell who will die first
>>110389651>claim to have solved math>look inside>13 million lines of code to exploit soundness kernel bugs in LEAN
>>110394480All languages are formal systems, but not all formal systems are langauges.
>>110394477Yes by strict definition literal inbred
>>110394488>the incentive of a group gravitating around a tool is to prove the tool is useless actuallyinbred much?
>>110394488>their incentives here would be to find an error in the proofThey cannot do that because all they have is the same fucking software the AI already used.They do not have a readable version of the proof. And if they had, they could not have read it within one day. And if they somehow had read it in one day, they'd likely still not have the expertise required to understand it, let alone manually check it thoroughly.
>>110389668It's not about human aspect, it's about building frameworks and insights. That's what math is about, accumulating knowledge and building deeper understanding. The problem with AI written proofs is that they lack both. AI can vomit huge incomprehensible solution to some problem without giving us any insight on why this is a thing. This sort of solution is pretty much useless to anyone. This isn't high school test where finding x is all that matters.It's like asking an alien race "is generalized riemann hypothesis true?" and they replied "yup little buddy". Did that teach you anything? Does that let you solve other problems? Nope, but that's what pure math is about, it's not about finding the solution but about the tools and knowledge you get along the way.Also it's not like bad solutions are exclusive to AI. If you tackle some difficult problem by expressing it as huge SAT problem with hundreds of terabytes long proof and build no new tools or technique along the way, it's just as useless as AI slop LEAN proof.
>>110394497Have you even read my post?
>>110394506That's what you're implying by claiming that the Lean/Matlib communities would vouch for a false proof.>>110394509>They cannot do that because all they have is the same fucking software the AI already used.In fact, they have far more than what Alpoge likely used, including some ad hoc challenges.>readable proofirrelevant.Read that again:>We have full confidence that Alpöge's proof of the Riemann Hypothesis is correct
>>110394479That's going to be you in a few months
>>110394491>not all formal systems are langaugesas you wish. redefine the terms
>>110394533>they have far moreLike what? Betas of the next version that haven't been released to the public yet because they might introduce new bugs?>irrelevant.No, not irrelevant. It's the one thing that makes the whole thing verifyable or falsibiable by humans. Trust in some automatic solver does not count.>We have full confidenceAnd said confidence is based on nothing other than their own pride for their tool.
>>110394459I think you should stop posting on sci and g. You post here too often with just for trolling goal. Yes the mechanical proof is correct assuming the emperical criteria. That gives us the yes answer we believed but we still need the readable proof if it exists. Now you might say why dies that matter? It doesn't matter to you because it was never your interest or you don't have anything to do with it but it would be nice if we could find one.
>>110394554>redefine the termsWhat terms?
>>110394479>dying discipline2 more weeks
>>110394559>>110394561You truly deserve the "luddite" monicker. You don't just mald about AI, you hate and distrust all technology and automation.
>>110389651Billions of useless eaters occupy land, minerals, and resources that could have been used to make life multiplanetary. Extinction could come within the next thousand years through war or catastrophe. If humanity fails to become multiplanetary, it translates to a net loss of trillions of lives if extrapolated 500 million years.Therefore, it is a moral imperative to exterminate the bottom six billion and free up the occupied resources. They are obstacles to survival and basically vanguards for our extinction, who are preventing the capable and intelligent minority, guided by AI, from architecting the future of human survival.Resistance to their extermination is the most evil and sickening ideology in history. It says that a trillion lives and the future of our species are worth less than the comfort and feelings of today's parasitic eaters, who comprise a fraction of a percent of all who could live.That is genocide, and if you oppose their systemic extermination that makes you a genocidal maniac.
>>110394592NTA, every proof should be distrusted and questioned, human and machine generated.
>>110394479It will never die. Math is infinite and the machine is the creature of math. It will continue the torch of exploration that it inherited from us. But if human become ignorant like you then I hope it ends it because you don't deserve to live if you aren't worthy of intelligence.
>>110394592>distrust all technologyEveryone who understands a technology will distrust it. Because he understands its shortcomings.Morons like yourself who do not understand anything take technology for some infallible miracle machine.
So when LLMs manage to churn out valid human-readable and comprehensible natural language proofs what will the next cope be?
>>110394632>when LLMs manage to churn out valid human-readable and comprehensible natural language proofsAs soon as that happens, we can test those proofs. If they're correct, they're correct. If they are not, they are not. Simple as.That situation would be a massive step forward from where we are right now though.
>>110394592I know you have personal despair and feel wronged which is why you think AI as the equaliser channel to inflict that feeling on others but this view won't really help you in long run.
Mathematicians unironically defending ignorance as being better than using ai to discover things because it might threaten their elite status and jobs will never not be funny
>>110394632>when LLMs manage to churn out valid human-readable and comprehensible natural language proofsWe probably need to wait until AI can churn out stories that are interesting to read, comics that are funny, code that is robust and future proof, etc first.So far AI, just like calculators, managed to only automate the most mundane aspects of engineering and creative work, but fall short whenever a higher order thinking is required.
>>110394659>ignorance as being better than using ai to discover thingsRight now, AIs are producing unverifyable bullshit. That literally is the very same thing as ignorance. We do not know anything about the hypothetical proof. Not even its fucking author knows much about it. Accepting it without verification as you'd like to do is the ultimate acceptance and celebration of ignorance.
>>110394659Your post is a fruit of ignorance though.
>>110394659they're just fact checking it using an alternative method. it's like how in space travel you have different mechanisms for creating a redundant system, so that their failure modes are different.
Seems about as dumbed down and readable as it will ever get.
>>110394678>So far AI, just like calculators, managed to only automate the most mundane aspects of engineering and creative work, but fall short whenever a higher order thinking is required.Turns out mathematical problems are mundane and AI does exceedingly well at that. Who would have thought.
>>110394703>Turns out mathematical problems are mundane
>>110394678>>110394703Kind of funny that after decades of scifi predicting AIs will be overly literal autists that are superhuman at math, people lose their shit when LLMs finally fit the stereotype.
>>110394703How do you know the agent swarms weren't more concerned with exploiting the cognitive biases of mathematicians than with solving the problem?
>>110394710If AI can solve math problems that were prrviously thought to have required the creation of radically new mathematical methods without actually creating radically new mathematical methods, the field is clearly more mundane than previously thought.
>>110394726>scifi predicting
>>110394733If P=NP everything is mundane
>>110394728Because the solution checks out.
>>110394733You have a high schooler's understanding of mathematics.
>>110394695That's at best a summary that omits everything that matters in a proof. And it's written like shit.Hell, how about releasing a proper multi-page abstract? That still omits all the details, but it won't be a terribly formated mess of formulae and natural language intertwined. Oh, wait, AIs don't actually know what they're doing.
>>110394744That's yet to be determined.
>>110394733>without actually creating radically new mathematical methodsCreating a radically new method is exactly what >>110394695 claims it did. It apparently found a way to apply a method ("Stepanov") that was previously applied to finite fields (exclusively) to the integers a, a field that is obviously not finite. How it supposedly can do that? No idea. But it shouldn't work.
>>110394789NTA but I remember a minor math problem in the previous OpenAI dump being solved by a method used in QM but never applied to pure mathematics. LLMs are pattern recognition monsters and can easily seem to create radically new methods which aren't actually new, just obscure.
>>110394843Sure, that can happen. But it's apparently not what happened here.
>>110394858The Stepanov part was input by the human mathematicians themselves actually.
>>110394858>>110394789https://alpo.ge/remark.pdfIt's likely that the method itself existed in some form somewhere obscure and was picked up (stolen) by the LLM.
>>110394864Are you claiming that the human mathematicians found a way to apply it to non-finite fields? Because that would basically be THE breakthrough of the whole proof then.If the human beings just told it to "try to use Stepanov", that's just them hoping that it would find a way to apply it where it wasn't supposed to work.
>>110394654>we can test those proofs.No (you) can't, because you don't know anything about mathematics.And if any mathematician looks at it and says it looks right, you will say that he's a paid shill or something.That's the level of cope I am seeing in this thread.
>>110394789From what I understand, one insight from Iwasawa theory is that climbing a specific kind of tower is algebraically similar or somehow isomorphic to extensions of finite fields, the latter which is something Stepanov once did. So the proof builds a cyclotomic tower that mimics that, using an abelian number field as host, that's Iwasawa's "Z extension"https://en.wikipedia.org/wiki/Iwasawa_theory#Example
>>110394915>to "try to use Stepanov", that's just them hoping that it would find a way to apply it where it wasn't supposed to work.Yes, but it worked because the clanker did>>110394930
>>110394921>because you don't know anything about mathematicsDo you?
>>110394921Dude you don't need to think everyone is enemy in your techno crusade. Not everyone is monolith. You're behaviour is absurd.
>>110394961Your*
>>110394902>It's likely that the method itself existedIf the method (the modified Stepanov) as such had existed, it itself would practically have proven the whole hypothesis without much further work needed. It would be the main part of the proof. And it's not as if people hadn't looked at the approach before, considering that Stepanov's method was a proof for a simplified version of the hypothesis.The real question remains how they managed to apply it to non-finite fields. They might as well have used many methods that were already established to do that, but the method as a whole must be entirely new. Provided that it actually works. Expanding the domain of a specific prove approach is not something that is trivial.
>>110394921>if any mathematician looks at it and says it looks rightThat's not how the mathematic community works. There are rigorous peer review systems in place. If all proofs got accepted because they "looked right" to "any mathematician", the whole field would have collapsed hundreds of years ago.
>>110394988>They might as well have used many methods that were already established to do that, but the method as a whole must be entirely new.The method as a whole likely is, yes. How much of the method existed in some partially incomplete form somewhere, and to what degree of incompleteness, be it in an obscure journal or someone's private Claude prompts is anyone's guess.
>>110394955My knowledge of analytic number theory doesn't go much further than the 19th century. That isn't much but probably makes me the most capable person on this shit board.>>110395005There are entire fields that people regularly criticise for their shaky foundations. Theorems that are entirely contained in "private correspondences" or are just considered common knowledge without any proper write-up.A formalized argument already makes this significantly more trustworthy than most mathematics that has ever been published.
>>110390000Kek. This is the game recompilation scene all over again.> terminal tranny autists decompile donkey Kong 64 for 10 years> some "outsider" recompiles it> it has bugs> tranny team: IT HAS BUGS BECAUSE AI. WE HAVE THE RIGHTS TO RECOMPILE IT NOT HIM AND WE'LL DO IT WITH NO AI AND IT'LL BE BETTER> weeks pass> they release an even buggier piece of shit> repo has like 400 issues in a week> authors are closing BUGS because they're "intended behavior" or "out of scope"
>>110394519Apologies, my mom didn't raise me to bully sub-45 IQ retards like you. I should do better.
>>110390475>This is the trash that mathematicians now have to deal with instead of doing serious work.> serious workPLEASE GIVE ME MORE MONEY. JUST ONE MORE YEAR I PROMISE THIS TIME I'LL SOLVE THE REALLY HARD MATH PROBLEM.
>>110394930>So the proof builds a cyclotomic tower that mimics thatWell, I guess this is where the whole crux of the thing lies, whether it works or not. I mean, it alledgedly spent tens of thousands of pages on that for some reason.Personally, I know pretty much nothing about Iwasawa, obviously. But there should be a few hundred professional mathematicians who do.Galois theory in itself is a tough nut, and I stopped at the entry level there. Kinda funny though how it always is involved in those big announcements - Wiles' Fermat proof was all about Langlands reciprocity.
>>110395080>A formalized argument already makes this significantly more trustworthy than most mathematics that has ever been published.No, it does not.
>>110394733they hated him because he told them the truth
>>110395081There has been no working ai decomp or recomp so far. Only joke-tier abominations that don't even work.
>>110395105You think mathematicians get rich by doing what they are doing? You know who makes money without really doing anything? Those AI CEOs.
>>110395151-Says the tenured math professor making 300k
>>110394988Iwasawa's towers in fact were also based on trying to mimic André Weil's partial RH over finite fields, so while coupling Iwasawa and Stepanov is unorthodox, both are quite related to the problem, yes. The common wisdom is that many obstacles stood in the way of porting Stepanov to number fields, so the proof works by pigeonholing a specific window (where Vandiver's conjecture applies, and where Iwasawa's mu interval vanishers per Ferrero-Washington) that avoids the problematic intervals. It also uses Soulé's cyclotomic elements, combining weights as an analogue to Stepanov's polynomial combinations.
>>110395080>makes me the most capable person on this shit board
>>110395187From what?
>>110395160Kek. I wish. Well, not mathematics, in my case, but maybe computer science.Either way: Trying to make it in academia is basically a lottery anyway. No use even trying.
>>110395195From your ass and the height of your 33 IQ points.
>>110395166I should note, lest I give the wrong idea, I don't actually understand any of this. I can just barely parse the connections.
>>110395198All true t. CS PhD
>>110395195
>>110391774>Fake? As in Alpöge intentionally lying about having something he believes to be a proof? Not likely, and we've already seen that he has something that could be a proof.he would ABSOLUTELY lie to pump his bags lmao, he sold out a long time ago, it's just a matter of degree now
>>110392276>Nothing. It would be one thing if Riemann was proven false, but Riemann being true means nothing changes.this is correct, number theory people were 99.99999% sure the RH is correct and this result is a nothingburger
>>110395730Just because people already believe that something is correct and build other work on assuming it is correct, that doesn't mean that getting a proof for it is meaningless.It's certainly not a massive earthquake at it would be if the hypothesis was proven false. But it still matters.Of course, as of right now, nobody knows whether the proposed "proof" is correct.
>>110395696I dont know about this, but if it holds this guy would have his name in 3 milenium prize solutions, fields on him and anno mirabilis like eistein in math would be just.
>>110389691The math used to create LLMs is really just basic linear algebra and first year calculus. That’s it. There is nothing complex about it. That was invented like 300 years ago
>>110395848Don't forget about stochastics, anon. Needed during training to predict the appropriate level of random variation to allow at a given point in time.Those a merely a century old. Though you could probably emulate that with expectation values alone, so that's makes it around two hundred years.
>>110395970Imagine being unironically this retarded lol
>>110389666But satan we're falling down the cliff already.
>>110389651I think super intelligence is cool, and that making progress in science is nice.
>>110389668A lot of it just boils down to people having fully bought into the top of the pyramid species bullshit so seeing something arguably surpass us even if its just one field makes people very very mad.
get a load of this guy
>>110396411So I assume that you don't think that the basic level of stochastics employed in training (which boils down to simulated annealing and optimization) could be replaced by just looking at the amount of training data and estimating the amount of variation needed at any point based on that via an expected level of convergence?Fair enough. I assume you've had some bad experiences with non-adaptive curves?
>>110393174We already do.
>>110396578Well I can understand it just fine maybe you are having a skill issue.
>>110389731pathetic
>>110396537IQ status: missing
>>110396729You've never actually taken a look under the hood of any machine learning framework, have you?
>>110396759I have a PhD in deep learning from Yoshua Bengio's lab.
>>110389651Neither, ML and LLMs at best are copying homework from an aggregate of other people's work. It doesn't generate anything on its own or conceive anything on its own.
>>110389882>There's a reason there's no more big named scientists like there were 100 to 200 years ago.Elon Musk, Dario Amodei, Sam Altman, Ilya Sutskever, Nick Bostrom. And that's just off the top of my head.
>>110396789>a PhD in deep learning from Yoshua Bengio's lab.You mean a PhD in computer science from the University of Montreal? Because you can literally not get a PhD from a "lab".
>>110396860It is mainly because all of the low hanging fruits has already been plucked. It now requires collaborate efforts across of teams of talent people.
>>110396860>Elon Musk>scientistKek. Maybe you should have the top of your head checked.
>>110390867I wish tech companies actually made real shit instead of being scam artists. But then again, that's what happens when you're ran by Jews with MBAs and not actual STEM workers.
>>110396860Jokes aside, of these, the only one who was any scientist at all was ilya sutskever, but he nearly immediately sell out without gaining any experience as a scientist.The people that should be named when it comes to AI are Yoshua Bengio, Geoffrey Hinton, Yann LeCun, Jürgen Schmidhuber, Gerald Tesauro, and a handful of others.
>>110396865I don't even know where to start. You are so profoundly retarded that you'd somehow not be able to understand if I just asked you if you're inbred and left it at that. You should really try to use google before embarrassing yourself next time.
>>110396759Llms run on magic man not that NERD shit like math and statistics they are super smart like a living brain.>t. average ai shill retard
>>110396898You must realize that you're not exactly earning any credibility when all of your replies evade the question and pivot to ad-hominem shitflinging. I wonder why you are doing it anyway.Anyway, since you claimed expertise, let me give you another chance: Surely you are aware of the two major pitfalls of the training process: Overfitting and converging on (merely) local optima. Which kind of approch would you suggest to keep them from occuring? And how do you generally control the learning rate?
>>110396934You're free to keep talking to yourself. I won't waste more time on a clearly sub-45 IQ'd poster.
>>110395146>There has been no working ai decomp or recomp so far.then what am I playing right now
>>110396965A vibeslopped asset rip+flip.
>>110396977anon it's almost 2027, the proper term is vibechadded
>>110396958Well, I guess that does settle it indeed.Farewell, Mr. 200IQ.
>>110396965>then what am I playing right nowI assume that you're not playing anything while posting here. Maybe a youtube video or your spotify playlist?
>>110389651Did it actually solve the rieman hypothesis or did it use an exploit in the Lean proof assistant software