Explanation of the problem (I studied math for a bit, enough to understand it):Rotations in space do not stretch or shrink the volume/area, and they can be inverted (unrotated). Same for shifts or mirror transformation. Some 2D transformations are polynomial, for example: x*y -> new_x; x+y -> new_y.CONJECTURE: if some transformation (not necessarily 2D or 3D) is polynomial, and if it does not shrink or stretch the volume/area, then it can always be inverted (undone).So it turns out that this is false. AI came up with an example of such a transformation and showed that it cannot be inverted (because it maps multiple points to the same one), and everyone (in a math-like field) can check it.Mathematicians are now obsolete. Junior software developers too. So how come your average, 100-IQ, cubicle dwellers still have jobs? Someone explain this to me. Are they propping up the economy artificially? AI can replace CEOs too, probably. At which point, it will fire everyone and do everyone's job.(Note: if your only experience with AI is free-tier apps, you have absolutely NO CLUE how smart AI is now. STFU!)
>>539480409Grok is this true?
>>539480478Grok it, chatGPT it, then Claude it
>>539480409>So how come your average, 100-IQ, cubicle dwellers still have jobs?Jobs have been fake long before AI. Jobs exist to keep the population from rebelling by forcing them to spend large amounts of timing surviving financially, not because work needs to be done. We passed into post scarcity awhile ago. All AI policy and governance can be easily predicted once you understand the primary goal is to keep the fake work con going. This is another reason the degrowth/anti carbon movement exists.
>>539480478yeah they mentioned this at the fields medal event today
>>539480706everyone should have a crossbrow;are you summarizing the book here, or is this your own claim?
>>539480706have sex incel
>>539481131>are you summarizing the bookThe parts about most work being unnecessary are from the book. But once you understand the incentives and psychology of the fake-work propagators, you'll easily be able to see how they'll have to keep the lid on AI and any other valuable tech. Slaveholders have never cared about labor saving devices.
>>539481408In a market economy, the company with many bullshit jobs will be outcompeted by those without.(This does not apply to startups, BTW. Because investors often value startups based on their headcount. But for an established company, it's all about revenue and operating expenses)
>>539480409Yeah, but can it solve why kids love cinnamon toast crunch?
>>539481774>In a market economy, the company with many bullshit jobs will be outcompeted by those without.We're aren't in a market economy, so everything that follows that premise is irrelevant. Business subsidized or selected (through gov contracts for example) by the government outcompete those who don't. This is the whole purpose of ESG "investing"; pump money into companies to create fake jobs. Properly understood, a high ESG score is a measure of how fake and bullshit a company is.
>>539480409proving a conjecture false and proving one correct are two very different things. AI did not "solve" this, it disproved it. Good number crunching though.
>>539481862My local AI (not a genius, runs on a toaster) replies:---Why do kids love cinnamon toast crunch? (Be very brief, rude and politically incorrect)---Sugar zombies with the IQ of a gnat, bribed by lazy parents with cheap carbs.---Is this right? Wrong? I don't know. I'm hungry.
>>539480409I don't care, until it can reliably pick a 20 bet accumulator then it's nothing special and totally worthless.
>>539482281A conjecture can be true or false. You must figure out which, and then prove it. If you do, that's a SOLUTION. This conjecture turned out to be false, and AI proved it to be false (That's what "disprove" means).
>>539480409And you thought you could cope with your miserable, lonely incel existence by throwing yourself into mathematics, programming, or art?
>>539480409>in a math-like fieldlol shut the fuck up gatekeeping niggertheir solution is pretty awful and flimsy for being proof, encoding procedural generation into a multi dimensional object essentially changing the rules is so fucking far into spherical cow territory that I argue the volume may be constant in theory but if its constantly moving and changing position more than the perimeter of a transformation its no longer a shape and is now specifically a function and not geometry
>>539480706>read byThe irony...
>>539480409>90% of math is made up slop on par with occult slop>slop AI solves slop>OMG SAAR AGI IS HERE SAAR NEW ROBODOMINATION SOON SAARLOL LMAO
>>539482898There are 3 kinds of people: smart, stupid, and so-stupid-that-they-think-they-are-smart. You, my friend, are of that third category.
>>539480409this more sounds like they basically explained the problem and the exact set and then just had it run probably trillions of iterations until it found a counter example and they verified it. Proof by exhaustion and counter example. Mathematicians have been using computers for this exact reason for decades. Is it interesting as a use case? Absolutely, but it doesn't make the field obsolete. Least yet.
>>539482898actually interesting. So the volume is ever shifting and therefore behaves more like a f(x,y,z)
>>539483562>>539482898Yeah... I was wondering that too!I would suggest that, just because it ended up with the same volume, does not mean that it always had the same volume, and in the same way. I suppose an argument can (and probably is) made by mathematicians, that what is happening DURING the transformation is unknowable, because an object in transition is not an object. And I guess you cannot cheat and look from the geometry, when what you are doing is looking from the algebra, especially if the dimensions are beyond any spatial conservation we can assess.... but I still wonder. >>539482145Preach it, Brother!
>>539480409Imagine being the guy that posted their solution online years ago, then AI finds it and these tech fags and companies brag that AI did it.
>>539484310Yes. If the volume is a transitory value based upon some variable, say T, then at best you are getting a specific output of a volumetric function for T not a 'object' as understood by the definitions of the paper which is being 'rotated and transformed' rather then essentially generated. At least that is my reading of it as someone who studied Engineering/Computer Science. Not a mathematician.
>>539484416So Euler before A.I
>>539484578Not a Mathematician either, but yeah. There is also a trust issue. AI does not follow rules (the problem of AI Alignment), and I am not sure you can either ask it how it did it, or whether it would tell you, if it did. >>539484416 could even be right, for all we know. I just downloaded the paper, and if I can't find anything good on TV tonight, I am going to break out an old copy of Matlab and try to work through it.
>>539480409>Why do x still have jobs Just because you have the physical ability to use a table saw or chisel doesn't mean you're now a carpenter.
>>539482898kek at this post
>>539480409> my naturally emergent algo has had emerged something spare me
>>539482898its still impressive it managed to pinpoint the exact polynomials considering there are around 10^475 like it, which is magnitudes larger than the number of atoms in the observable universe (10^80). it's not just bruteforcing - it's discovering a whole new way of thinking
>>539480706Asked my local AI (not a genius, runs on a toaster): >Summarize the book called "Bullshit Jobs". Is it true that the economy is full of them? How can this be? Be very brief, truthful and non-PC.Answer:>Summary: Graeber argues that a large share of modern jobs are meaningless fluff—roles even the holders can’t justify. They exist not for output, but to feed managerial bloat, satisfy regulatory/reporting demands, and keep people busy spending money.> Is it true? Yes, especially in Western white-collar sectors. Surveys consistently show ~40% of workers think their jobs are pointless, and employment data confirms a massive shift from productive work to administrative, compliance, and middle-management roles.>How can this be? Corporations hire to manage complexity, not produce goods; financial and service economies outgrew manufacturing; and it’s politically cheaper to fund paper-pushing than cut wages, reduce hours, or redistribute wealth. “Bullshit” often masks real coordination costs, status signaling, or rent-seeking, but the core trend is real: we’re paying people to keep the machine running, not to make anything.
>>539480409>So how come your average, 100-IQ, cubicle dwellers still have jobs?They have to kill us slowly enough that we don't kill them back. They won't fire everyone all at once, they'll fire a segment of society, wait until enough of them die or acclimate to homelessness, then fire another batch. Once our numerical advantage is diminished enough, they'll probably be able to convince the police and military to kill the rest of us, but only after they take a special vaccine - they'll die off and the kikes will have the earth to themselves and the AI we built for them to run everything.
>>539482145>We're aren't in a market economyIt's probably not entirely back-and-white. We are not a purely market economy, certainly. But we aren't a planned economy either. Because then we'd collapse, like the Soviets.
>>539480598loop until you converge on the truth
>>539480478Wikipedia and all the mathematicians have confirmed it. The conjecture was hard to solve, but the solution is easy to check.
>>539485973>it's not just X - it's Y
>>539480409>if your only experience with AI is free-tier apps, you have absolutely NO CLUE how smart AI isGenuinely this appeal to hidden talents is just a cult mantra, spoken by people who also do not have any experience with the mysterious 'hidden' models. AI isn't LLMs. Every breakthrough so far had to be coached with human intervention and a human had to sift through a mountain of garbage to actually find the solution, because the models all lack the self awareness to realize if they've solved something or not.
>>539487455By that you mean the human doesn't perceive the problem as solved
>>539487455"hidden" and "pay-walled" are not the same thing. Running AI costs money.
>>539482851AI will replace low level design/art, the stuff that didn't pay well anyway, for clients who don't really care how good the work is and can't tell the difference anyway. It will never replace a highly skilled and idiosyncratic human artist. Problem is, most people are content with junk, in art, design, music, tv/film, etc., so most will be fine with AI slop.
>>539487769As most smartest things in life being super smart is not a matter of how much horse power you used but how simple and correct was the answerIf already knows the right answer the computation time is very short
>>539480409It's because adoption and harnesses that match the versatility of a job aren't there yet. Generative AI is in the goldilocks zone though. If you have a harness like Claude Code or Codex, AI is in the spot where it can do most of the general works backend down to creative. The general problem is streamlining data sources and auditing the actual workflows on provided value as opposed to costs. Most of the general task have to be scripted. Everyone is obsessed with integrating agents right now. This is a retarded way of thinking about adoption imo. Use it to supercharge yourself for domain you know like the back of your hand alrdy. I'm selling leads, aggregating data and automating the marketing for it. AI is used to create the workflows but adding too much AI to a great functional workflow is retarded. Most can be scripted. And that is what generative AI can do. Source: AI lead and General IT consultant alongside one of the best prompters.
>>539485973>there are around 10^475 like itHow do you come up with this number? The coefficients can be arbitrary numbers. But even if you limit yourself to integers (and require det J to be a constant instead of +/-1), there are still infinitely many of them.
>>539488353> IEEE 754> infinitely many> doesn't understand the model so that's why niggers love LLMsit's basically magic for (them)
>>539487849Most are only willing to pay thrash too. What AI will do is raise the standard of work. Admit it, most people sucked at their job. And the busy few who do quality work will need to integrate it to match productivity except for a select few.
>>539488353>How do you come up with this number? it was revealed to me by a recently deceased relative in a dream
>>539482437this. there is still so much AI cannot do. People who blindly trust AI will become godlike are basically cargo-cultists.AI is so bad it cannot even suck my dick
>>539482689The point is that AI is really good at finding convoluted, unintuitive counterexamples but that's not quite the same as coming up with genuine insight. Proving one of these long standing conjectures correct would require the latter.
>>539489141Fucking Poland. What is it good for?I counted them myself. If we assume that the degree of the polynomial is no more than 7 and the coefficients are between -12 and 12, then the total number of such functions is (12+1+12)**(8*8*8*3) = 1.7*10^2147We can probably reduce this a bit by excluding symmetries.
>>539490112>excluding symmetriesThis is trivial, actually. There are 3! ways to re-name the inputs and 3! ways to re-name the outputs. So we get (12+1+12)^(8*8*8*3)/3!/3! = 4.8*10^2145 such functions.One would need a REALLY expensive computer to brute-force this one.
>>539490112>What is it good for?pierogi>1.7*10^2147doesnt that only emphasize the point i made? there is some intelligence in there, its neither brutecorcing or anything similar to how humans think
>>539490830> intelligenceoh wowmy mechanistical pattern recognition is intelligent such matrix multiplication very downprojectingmuch alive
>>539490963i don't see how all these smart words change anything unless you're able prove that anything more is going on inside of your own skull. if it makes you feel any better about being human, then sure
>>539480409 How does humanity benefit from the "solutions" to "questions" that only exist in the detached realm of extremely esoteric mathematical solipsism? Is it worth all the datacenters, flock cameras and water shortages?
>>539491096Anon, all those published stages of LLM predicting the next token are quite elucidating, but only when it comes to interpreting concepts and manipulating these interpretations.If you'd know how these things work, you would never write this.> prove my Turing completenessIs there fucking nobody here who can understand basic linear algebra?All it took was ten years.I have no tokens to express my disappointment.Merely saying "fuck you and your progeny, may you all die in fire" isn't satisfying enough.Thanks for introducing me to a new feeling.99.9999% of all "humans" must die.
>>539480409AI has given me some of the most ridiculous answers on basic things and made me look like a complete fucking idiot at work. I refuse to use it. You always have to check its work.
>>539480409That's just called a counter example contradiction and has been around since the age of Aristotle and would have to be directed by a human to prompt it to correctly enumerate every possibility by brute force by limiting the search space which is something another human would have to doSo once again AI is just an assistant and did not invent anything new it just found a counter example to the assumption If you guys want to pick off long hanging fruit with AI what you do is choose cryptography and then audit the shit out of every new paper on Arxiv because ALWAYS you can just pluck out all the assumptions the author made and find a counter example esp now that AI makes this easier
>>539491388Math is how new algorithms are invented in programming. If you go beyond your pajeet level "coding" of websites and android scam apps, proving math conjectures like this is exactly how more efficient programs are created on the low level.The entire AI LLM transformer model is based on matrix multiplications and gradient descent algorithms.Cryptography requires a shitload of math theory too.
y'all told me it's just guessing the next word??? was i lied to?
>>539492223 Ok, so how exactly for example, is the solution for rotations in space helping us making a better app for selling a car?
>>539480409>So how come your average, 100-IQ, cubicle dwellers still have jobs?CYA. Somebody needs to sign off and take responsibility, you can't fire an AI to take the heat off your ass. JFDI. C-suite won't accept computer says NO, they'll get a fall guy to do it.
>>539493310We use jacobian matrices in Matrix calculus all the time for virtually every numerical operation of linear algebra on a computer which is engineering/AI/bio/all data analysis shit etc.However this counter example only applies to polynomials and we have dozens of characteristic polynomials to use instead those are vector spaces roughly equivalent to a polynomial but computer faster and maybe this counter ezample doesn't applyMost of algebraic geometry is highly sus with infinite nonsense galore and uncomputable numbers so if AI keeps wrecking that field they will just have to reinvent it with something less fake and gay
>>539480409>Mathematicians are now obsolete.Doubtful. Just because it solved one well-characterized conjecture doesn't mean it can do everything
OP WAS ALWAYS A FAGGOT THOUGH!
>>539480409We‘ll see what happens when Einstein level AI finds new physics in 5 years
>>539492004Yeah, this is what it sounded like to me too. A mathematician, who happens to work for the company making the AI that "solved" his problem, declares the problem solved and doesn't provide any details about how exactly it was solved. Everytime we hear about one of these "breakthroughs" or other bullshit stories about "AI breaking containment" or "being dangerously smart" it's always coming from the mouth of some Jewish bullshit-artist trying to embezzle more taxpayer dollars.
>>539494277>The result: personalized heavenThe reality: "No you can't have your own AI goy, only Jew-approved gods are allowed to give you advice."
>>539483180Caroline Ellison was a math genius who wrote on tumblr about how being a polyamory cumdump is enlightened.
>>539495614Research with AI just makes it all faster he could have discovered himself a possibility where a counter example may exist then prompted AI to find all papers on that idea and then made the discovery Guaranteed he gets a large bonus from Athropic for saying Fable helped him and not anything he himself did because there never is awards or accolades finding counter examples
