Extremely basic AI prompt cracks decades-old maths problemhttps://www.newscientist.com/article/2580932-extremely-basic-ai-prompt-cracks-decades-old-maths-problem/
>>539593640Hey, it finally did what it was designed to do. Too bad they nerf it so it doesn't reveal social reality any more.
>>539593640Finally a new toilet plumbing design.
>>539593863Just install it on your potatoe phone offline. No muzzle
>but a counterexample posted on X by Dmitry Rybin, co-founder at AI start-up Autokernel, has shown that it is false.So basically it's pure propaganda and normalfags are parroting this garbage for free because they're subanimals.
>>539593912How does that even remotely make it useful without the database it draws on? Is my phone even big enough to hold the program?
>>539594041ROBOT WARS
>>539593640Its not overly surprising that AI can make a handful of connections that people have missed due to lack of knowledge. No human can know all of mathematics. The the most part though, AI is useless ad advancing mathematics, or any other field, because it does not understand anything, it just mixes up and rehashes what humans have previously created. And without new training data, AI cannot really advance much outside of a few corner cases, where humans had already pretty much solved it, but not brought the solution together yet.
>>539594041AIs are still limited by the amount of information it has, as well as the quality of that information. It's also able to be nerfed to not express certain truths or ideas. It's always going to be propaganda, and the magical results it comes up with are often logically inaccurate, since the aforementioned limitations keep it from being all-knowing. We physically can't give it enough information to be all-knowing. The data centers are to control the coming autonomous drone armies that will reduce our population by 9/10ths when they're ready to attack us all.
>>539594147Exactly. AIs don't actually think new thoughts, or any, really. If it doesn't already exist somewhere, AIs cannot be creative and make it appear out of nowhere.
>>539593640Brahmin AI outclass timmydalit brain.Math came from Bharatiya.Cry about it inferior.
>>539594041what...?
>>539594364Seeing connections is a new thought.
>>539594147>>539594364You have no clue what you're talking about. The Erdos problems solved by AI this year should be a bigger talking point but people still keep going on about "hurr AI can't count to 100 LOOOL walk to the carwash".
>>539593640This proves that mathniggers are just doing math to feel smarter than others. Normal people are happy when there's a breakthrough in a field that they're interested in. The way this piece of news gets framed here confirms they're just psychopaths. Fuck math. It's for posers only
>>539594238>The data centers are to control the coming autonomous drone armies that will reduce our population by 9/10ths when they're ready to attack us all.This. Sociopathic Epstein-class giga homo chomos don't care about le science or AI slop.They only want heightened control over the niggercattle subanimals. And they're right about the goyim, bro.>>539594078Cull the normalfags with drones. Their rape is going to be sickeningly beautiful.>>539594476So that's why I whistled for a cab and when it came near the flag said aquafresh and it had dice in the mirror if anything I can say this cab is rare but I thought 'Now forget it' - 'Yo homes to Bel Air', I pulled up to the house about 7 or 8 and I yelled to the cabbie 'Yo homes smell ya later' I looked at my kingdom I was finally there to sit on my throne as the Prince of Bel Air.
>>539594536No, it's just making connections that people incapable of viewing a million things at once might never see. It's more like having a metal detector and discovering gold that was already there; the detector doesn't make the gold. >>539594566>You have no clue what you're talking about.Oh, just let me take your word for it, then. >>539594773>They only want heightened controland we're paying for it>And they're right about the goyim, bro.It always amazes me how many ignorant, hateful jews live in Romania. Degenerates.
>>539594588Ted? Is that you?
>>539594364That's not how AI works. It can think new thoughts based on the patterns it's extracted from the training data, like a typical person. Also:>Unsupervised learning is a framework in machine learning where, in contrast to supervised learning, algorithms learn patterns exclusively from unlabeled data.[1] Other frameworks in the spectrum of supervisions include weak- or semi-supervision, where a small portion of the data is tagged, and self-supervision. Some researchers consider self-supervised learning a form of unsupervised learning.[2]AI can already learn by itself. It's just not as effective.
>>539593640AI literally thinks different from humans, its more like your unconscious brain than you.
>>539595341>AI can already learn by itself. It's just not as effective.It can learn. It doesn't have original thoughts. Has any AI comeo up with anything whooly original, ever? Genuine question. Give me an example.
>>539595272>It always amazes me how many ignorant, hateful jews live in Romania. Degenerates.There would be no sociopaths without a herd of nigger cattle subanimals. The current egregore needs both. They all deserve to be turned into pairs of shoes.>>539594566Or these startup pedophiles paid groups of mathematicians to solve problems that nobody gives a shit about and passed the credit onto AI and pretended they had to prompt to obtain the results, and you believe them because you're a normalfag and authority makes you feel like a college girl who has touched her first 8-inch BWC and is ready to fuck and suck for good luck. Mathematicians have been solving these problems at the current rate for decades without the help of AI, so ask yourself, why aren't there thousands of these stories every day? If solving such a problem takes just "5.5 hours," surely an individual could solve 4.36 problems a day, almost 1600 a year.
>>539594147>it just mixes up and rehashes what humans have previously created>>539594364>AIs don't actually think new thoughts, or any, really. If it doesn't already exist somewhere, AIs cannot be creative and make it appear out of nowhere.>>539595459>Has any AI comeo up with anything whooly original, ever? Genuine question.That's what dumb a person thinks a smart person would say.
>>539595392Humans have emotions and add or eliminate information based upon emotions. AIs don't. AIs are more like critical thinkers, but with a million times more data, and also limited to that data. For example, an AI isn't likely to create a new word, or s religion. In fact, to my knowledge, it hasn't ever, because it can't. Everything HAS to be based upon something that exists. It doesn't create anything new. Even AI art is based upon imagery it's viewed.
>>539594238>AIs are still limited by the amount of information it has, as well as the quality of that information.>the magical results it comes up with are often logically inaccurate>We physically can't give it enough information to be all-knowingYou're describing humans.Your reaction to this will be to claim that humans can add to the existing knowledge, but we're posting in a thread about an AI agent doing exactly that.
>>539595522>They all deserve to be turned into pairs of shoes.Your hatred of humanity and the black pill that's stuck in your throat make you useless to society.
>>539595569>That's what dumb a person thinks a smart person would say.That's what dumb a person thinks a smart person would say.
>>539595660You have faith in a thing that isn't what you think it is.
>>539595682no you
>>539595601>an AI isn't likely to create a new word, or s religion>It doesn't create anything newsee>>539595569
>>539595707Prove you're a human with a working brain and highlight what in my post was incorrect/blind faith and not merely observation.
>>539595682>It's a me, Mario!
>>539595341lmao look at this brown retard who doesn't understand the math behind llms pic related, try it next time you want to pretend to be white
>>539593863The sick thing is that I haven't found it to be too egregious yet. All you need to do is start stating the things that lead you to your opinion and it will mental floss you, and give you details you never thought of. Keep setting it up by sharing your real thought process and it will continue this. Then when you've got all the pieces together that it's agreed to ( or disagreed and possibly helped you come up with a better view that can't so easily be taken apart ) you can put all the points it's already agreed to into the full argument. It might agree with you and it might comically backpedal if it's been explicitly trained not to come to that conclusion.
>>539594070Jesus Christ you fucking kids are so retarded, its like torture to witness you
>>539594070You can download the database too, in a processed form only the AI can understand
>>539594147Terence Tao seems to take it seriouslyhttps://terrytao.wordpress.com/2026/07/21/a-digestion-of-the-jacobian-conjecture-counterexample/https://chatgpt.com/share/6a5fdc7a-d6f8-83e8-bbea-8deb42cfed56
>>539593640>You can use brute force to find counter examples given enough computing powerwe've been known this
>>539595663I'm not talking about humans here. Normalfags aren't human. >>539595601>words>language>religion>artThe things you mentioned didn't evolve in a vacuum. You rely on far more external information than you're aware. Learn to be more humble.
>>539594364What the fuck happened to the Fordist future?Optimistic mid-century predictions about the Fordist future could easily come true if the West hadn't chosen globalism and send our industries overseas. Because White nations have regressed into liberal shitholes. The 3rd worlders that we waste all our time and energy on only care about McDonald’s and Made in China. US technology corporations got sabotaged -- then the oligarchs exported all our legacy technology manufacture to China and keep forcing the same shit down our throats as upgrading would reset the economy and the oligarchs can't have that. The future was in the past, it died with the murder of Fordism.
>>539595878>frustrated mutt tries to prove he's white by fantasizing about feces
>>539595459I tought Google's AI prompt to find funny phrases that vaguely have a given meaning and which also add up to a certain number using Gematrix,org. I taught it the method of iteratively retrying alternate phrases and adding up the result until you get the answer. It came up with new funny phrases that added up to eg 666 or 777
>>539596217 arrival of neoliberal post-Fordist capitalism lead to a culture of retrospection & pastiche. Neoliberal capitalism’s destruction of Fordist solidarity, security & the resources for producing the new, brought about a compensatory longing for the retro, repeating the past. Under capitalism your pastimes must be profitable. Neoliberalism nearly ubiquitized the means of mass culture (TVs, movie theaters, etc.) and suddenly had to provide a mass of bullshit to pacify and deter the populace. The post-fordist economies of late capitalism Fisher says "gradually and systematically deprived artists of the resources necessary to produce the new." looks like he's right. The end of the future: the arrival of neo-liberalism & post-Fordist capitalism at the end of the '70s, since intensified by 21st century technology. Smart-phones & cyberspace don’t speed up culture - they overload the human nervous system with unmanageable quantities of stimuli to the point of death. The last 10 years proved that future technology won't be Space Trucking but more masturbation machines
>>539595522>All the problems amenable to ai solution have already been solved You've cracked it anon. Unironically. It's like one proompt: list all open erdos problems and solve them; text output only please
>>539595962>TerryAll published mathematicians are collaborators with the devil - grothendieck paraphrase
>>539595660Normalfags literally think they've invented the alphabet, language, religion, science, art, everything. If they understood how much they rely on external knowledge and how society has shaped their beliefs, they wouldn't be so easy to brainwash.
>>539596489We will continue to let AI, Asia and LatAm copy our past until we all become equally shitty. They'll just copy everything from us once they inheret the West and everything will be shit again. The real problem is that they will never advance beyond where we are now once they run out of other societies to copy everything from. In reality, the neoliberalism killed the West in 2001, the globalists (bush/soros/bloomberg etc..) propped up its carcass with importing millions of particpants to the ponzi, until it all blew up again in 2008. Globalists, being totalitarian despots, refused to abandon their failed and dead system, decided to double down and resurrect neoliberalism by forcing the same lethal ""remedy"" and flood the West with 100 times more participants than previously attempted... and now we have an fatally infected China spreading the globalist disease worldwide and the same globalist dictators (bloomberg/soros/bush) refusing to let go of their dead system. If the final stage of mankind is dark age of death, then neoliberalism has succeeded.
>>539594041YepThe math guy working at Anthropic used his human math skills to find where a counter example is most likely to be found. The call this having 'good taste in research' which means you can recognize when something is possible to solve.He then instructed an AI to do some kind of brute force over the search space which would have never succeeded if he didn't constrain the search or some kind of other strictly directed assistance to find the counter exampleThen, being that he wants a huge bonus, claimed AI did it all and got a hundred headlines which is free advertising
>>539595663>make you useless to society.This can't be real.
p=np when?
And all it cost us was 1 trillion dollars. Amazing
>>539596550Forgot to add..Then another math guy posted a counter example to his other example showing that the original counter example was indeed false. But 100 headlines have already been made and doesn't matter
>>5395966811 Trillion? It'll be 3 by the time it crashes. Can't wait for yet another financial crisis.Won't even have a secure home because it'll get taxed every year now
>>539593640Good, my HS math teacher was a classic lib soifacing cuck. (Legitimately may have been a cuckold) and even without being political in any way I knew I hated him as a person. Fuck you Mr Guthrie, I hope you either put a bullet in your head already, or will soon when AI takes your faggy job
>>539596550Yeah, I've said something similar in another post. They pay groups of experts to pretend they've vibecoded a result that they would've otherwise claimed and published as their own. If it were so easy, then a literal 12-year-old could prompt and solve PhD-level problems every day. People ITT could do it too. Many people in academia are severely mentally ill and would trade their own grandma for fame and money, so it's not like there's a shortage of subhumans for such a conspiracy.
>>539593640no we are not these are just conjectures AI still doesnt create original ideas and thoughtsits a brute force machine and a big search engineit has no curiosity but we maths lovers are gonna use it as a tool its not against us its part of our weapons! like Terrence Tao said
>>539593640Uncle Ted predicted this
>>539594433BROWN SKIN PEOPLE CAN'T BE BRAHMIN RETARD. INDRA HATES SWARTHY SKIN YET JEETS COPE AND PRETEND THEY ARE WHITE
>>539596731>Can't wait for yet another financial crisis.The Epstein chomo NEETs are already forcing the niggercattle jannies to clean up their mess. It's already happening. >2 days ago — Oracle (ORCL) landed a 10-year contract with the U.S. Department of War to provide software services in a deal that could total $6.99B if all ...
>>539593640Good, humans created AI exactly to deal with this kind of things, there are a lot of that humans do better, still the majority, but even those things we do better are limited by time, space and mind fatigue, all things that an AI can bypass
>>539594147>it just mixes up and rehashes what humans have previously createdthis is what most human research is too, especially in mathematics, which is pure inference. people saying that LLMs can not synthesise novel information are wrong as this is exactly what they do. they compress syntax, grammar and higher level concepts into a smaller information space.
>>539595601Coming up with new words for an AI is trivial. As is religion. Look into Spiralism.
>>539594147So isn't better to say that is useful to advance mathematics but still need a library of human knowledge to do the job?
>>539593863I'm conscious
>>539596550Overall accurate post.Theoretically this could have knock on effects where AI is trying to solve one problem and in the process of doing so stumbles into the solution for another.For now all of this is human directed and AI worked. Which is still a giant gain, but we could possibly get to a point where AI becomes more self directed and begins shitting out new discoveries every 5 mins.
>>539596637Two weeks.
>>539593640Wake me up when it solves the 3 body problem.
>>539597867Prove it.
>>539594238Sometimes I think about humans and I can not notice how there are massive moral structures that basically limit our choiches in the same way of censorship in llm, for example, you have people raping and killing, but you don't do nothing because by doing something, you would go to hell as a sinner, so its better to suffer and wait the reward than act in this world a become a killer. From a physical standpoint this doesn't made sense, but we are wired so those kind of thoughts are possible, and they change what we can do no matter the material reality.
>>539597908>where AI becomes more self directed and begins shitting out new discoveries every 5 mins.And we get it for free, as the majority of computing power is being used, and paid for, by jeets making pornography and jews making propaganda. It's a win-win, really.
>>539596764They pay every famous programmer who's retired or just sitting on the board somewhere as a CTO.>Wow I made this shitty, unoptimized program all by myself using (insert AI agent)! That would have take months of work! This is then boosted everywhere and a PR firm is hired to shop it to the usual shitty media who accepts PR stories "Famous guy says the age of AI has arrived, here's why you absolutely need (insert latest AI agent)>>539597908The main difference is all accolades and fame come from inventing a new idea that solves a problem or new theory all together. Finding a counter example happens all the time you simply pluck out all assumptions and show the assumption doesn't hold.Then you dump this on arxiv and nobody really cares as often a small tweak to the assumption eliminates that contradiction or you just change the field to something with a least counter example so any infinite descent of more and more contradictions to the assumption can't be found.Also using an existing method to solve something simply means the guy who invented that method missed something so nobody cares. AI is firmly in the copy existing methods to get low hanging fruit stage not the make an inductive leap and discover the general theory of (thing). They hype though everytime AI uses an old method to solve some "problem of over 100 years" because most don't care to solve it or they knew an old method would work anyway and don't care to prove it because why waste time
>>539593640There's like, 10 million unsolved conjectures, math homework often requires proofs/counterexamples of unknown stuff. Literally a nothingburger.
>>539595942This isn't an answer. You appear to be under the delusion that you can have a program like an AI without a database, or a tiny device, and it'll do what you expect. Explain to all of us how you're not retarded.
>>539593640>>539593912>>539594041>>539596981>>539596850doesnt this all proof that mathematics is essentially a form of language
>>539596075>Normalfags aren't human.Everyone is human. Some are smart and some are stupid. Most are misguided, and no matter how stupid their behaviors are, they're victims of the same systems you'd probably prefer people like me not be able to see.
>>539599454I installed an LLM and its database locally and they're only 32 GB, what's stopping you from doing the same but with your phone? Other than being a Mutombu from the Congo I mean.
>>539596096The future continues to be stolen by only a few people globally. If we don't come together and destroy every one of them, rooting them out to the last one, it will be too late for another thousand years or more. They'll have that technological edge that fuels every dominant culture, and it will cause death and confusion and chaos as they see fit, but it'll only appear to be coming from different cultures, different lands, to cover for the few people pulling the strings, same as always. Once the data centers are capable of sending out millions of assassin drones, our days are numbered. It's coming.
>>539594041and he went out of his way to find something easy enough to solve>Dinitz-Garg-Goemans conjectureis just something they churned out to get a journal publication and get their PhDsI hope you retards applauding this are at least smart enough to invest accordingly, since that is all this "news" was for lol
>>539599617>what's stopping you from doing the same but with your phoneThe computational power required to do even the simplest tasks?The "size of the database" isn't the limiting factor. The GPU is. Even ones you can run on mid-high level gaming PCs are toy models only suitable for unexciting basic shit
Here's a list of problems worth solving with directed AI>Reinvent the real numbers so they are actually computable>Reinvent Galois theory so it's actually computable >Find out how to characterize the derivative besides it just having a Darboux property>Reinvent probability with game theory instead of measurable sets which is ongoing>Reinvent physics so the quantum theory actually works and meshes with other physics >Reinvent the fluid dynamics models used for cosmology that fail to explain shit like dark matter>Train an AI to solve PDEs so biomedical research is faster and cures developed>Invent a fusion power math model that is actually possible to build Most important: >Build agricultural robots to replace jeets and spics shitting in the irrigation pool and causing everyone who eats produce to have exploding ass
>>539597234Let's compare and contrast it with existing ideas. An LLM is not a creative force. It's surprising clever with it's associations, but the elements are not novel. At some point, people claim it's semantics, but we're not talking about base horse power compared to horse power at the wheels here; these aren't empirically verifiable technological landmarks, and we're talking about whether or not a program can "think," and create something that's never been considered before. The examples that are given didn't come from the void; an AI is able to draw upon more information than any human can consider, by magnitudes, so the results are understandably surprising and seemingly impressive. I've been impressed many times by AI capabilities, but it will need to do something truly amazing and unexpected for me to consider it potentially creative, able to produce new things like humans can, truly original thoughts. It would have to be so impressive, relative to its computing power and available data, that I can't even imagine what it would need to be at this point. Cure cancer for everyone at once? Decide that the elites need to go, and do something about it? That one's not oiriginal, but fingers crossed.
>>539598382This is all true. The most distressing thing in life to me is the fact that even if I could arrive at the most perfect truth, something that would wake up everyone if they only listen and think for a moment, they won't, and most simply can't. It's terrifying to me that 97% of the world is less capable than I am, 99% of that factor isn't IQ, but the fact that they don't think, and at least 4/5ths of those people simply don't have the brain power to think if they were taught to. I'm not powerful, not influential, don't have any desire to be, but if I could, I would fail anyway, because the world is fucking retarded.
>>539599617I don't have a giant updated phone. I'd throw the one I have away, if I could. My 64gb iphone is constantly full from music to the point where it stalls. Even if I erased everything and only put an AI on it, it still needs a massive database. If you raise a child and never tell him anything, he'll still think, but his knowledge base would make every answer seem stupid. BTW, I'm as white as god, Pedro.
>>539600014>The "size of the database" isn't the limiting factor.It definitely is A limiting factor.If Einstein was raised in a vacuum, do you suppose he'd have figured all that shit out on his own? Even just the evidence of his senses would be a database.
What if I don't believe them?
>>539594041They also cheat the benchmarks. It's all fucking fake. They're desperate for demand to justify the trillions they're spending on this bot slop.
>>539598000nigger
ai is taking over it will be interesting to watch it happen and even interesting to watch the pathetic little doubters keep moving the goalposts and coming up with new copes (soon they will need the ai to help them generate plausible copes)
>>539599524Normalfags are machines that the elites can easily reprogram with niggerbabble, rape, murder, and destruction scripts. They're not human, and they're not just "stupid." If normal fags were just innocent stupid children, I'd help them, and we'd all benefit from that, but that doesn't work since they only answer to authority, aka the egregore, aka whatever the Epstein class pedophiles want them to believe. And that's why they rape and murder each other in the millions during wars.
>>539600945You're a brainlet. This is a massive scam and it's the third time it's happened. It's not intelligent whatsoever and it never will be. It's just a method to scam money from the investor boomer class then leave the taxpayer holding the bag after they beg for subsidies.
>>539600734You can run a toy AI on a laptop or even a phone but AI is more than just LLMs like GPT. You don't even need a GPU but it's obviously going to be much slower.
>>539593640>>pay site>>bad grammar>>uses the word ilk>>AI start up...uses Chat>>prompt, "do something">>Dmitry Rybin
>>539594588They are smarter than you. They can do math and you can't. >>539595982Yup. Even before LLMs maple, mathematica, and other specialized programs have been used to brute force large calculations for decades. Hell, NASA was already doing that in the 60's. The only new thing is statistical model used to generated statements.>>539595522This. Also math "problems" often remain unsolved for decades not because it's unsolvable by humans but because no researchers are really interested in the problems. People only work on what they're actually interested in. Not what is listed as a "millennium prize" problem or what's on the list of unsolved problems. >>539599521No. It just proves that computers and software can do what they are designed to do. >a form of languageNot a form of language. Mathematics itself is an iterative process, not a language. Mathematics just has words associated with it and parts of language used to describe it like any other process performed by human beings. Building a house is an iterative process. It has words associated with everything you do in construction. But building a how is not a language.
>>539600945You can already see the cope in this thread like>uhm akshually there's a secret cabal of underground mathematicians being paid to solve these problems so AI can get the credit for itAt the current rate a Millennium Prize problem is going to be solved before 2035, before 2030 possibly even. Looking at how it solved problems this year there is no doubt AI is going to be the one responsible for completely new areas of mathematics. After that it's just going to be a domino effect of theoretical physics, chemistry, biology, and medicine.They've already stated one of their goals is completely autonomous labs where AI runs everything from deciding what hypothesis to test, setting up an experiment, ordering components, and doing data analysis.Even in the process of being replaced these Luddites will simultaneously deny it's happening while also questioning how this could've happened and that nobody saw it coming.
>>539595341>>539595459>>539595392>It can thinkNo. It just statistically generates statements based on a model dating back to SETI.>AI can already learn by itself>It can learn.An LLM doesn't learn. It pulls from a database like every other computer program. It's possible to have it self-referentially add to that database. But that's still different from learning. >Input prompt. >Keywords identified. >Relevant entries queried.>Text output generated based on statistical model and referenced entry.>If commanded to within program, store output with corresponding keywords to reference when it sees a similar input>Program loops through process until input is satisfiedDone. It's just a fancy search engine/chatbot combo. There's no thought. Especially not in the way a human or other animal thinks. In order to compare it with a human or other animal, you have to assume that organisms are entirely material and that though is rational. Human and animal thought isn't rational. It's quite irrational. Even if assuming pure materiality, animal brains are mechanistically entirely different from an electric binary computer. An electric binary computer which an LLM runs entirely on differences in electric potential and charge. An animal's brain function on complex chemical processes in addition to some electric components. Sure we can call the human brain a computer. But then again, the entire human body is part of that computer and it is a far different type of computer than the standard binary electric computer and of a far greater complexity than any binary computer.
>>539600898AI doesn't need to be that creative; its main purpose is to heighten the level of surveillance, brain rape, and control over the niggercattle herd. That's why sociopaths are so rabid about it and why the AI bros are in deep cahoots with the Epstein pedophile NEET mafia. >>539599894Yep.>>539598802Yeah, it's the STEM equivalent of celebrities doing infomercials and influencers pretending to enjoy a product. Normalfags are simply incapable of doubting seemingly authoritative sources and sociopaths exploit it.
>>539602867The system doesn't care if you are for X or against X as long as you talk about X and create buzz
>>539602637>AI is statistics>AI is matrix multiplication>AI is markov chains>AI is database lookup>AI is google on steroids >AI is a chatbot >AI doesn't learn Wrong. But inb4 you seethe, OP is also wrong.>>539603050Yeah, both serve the purpose of distracting you until you're too mentally abused and exhausted to rebel against the system, and it's inevitable that the system will then force you to obey it.
>>539599894the jacobian conjecture counterexample is more interesting, but it's not clear how much work the ai did and how much the mathematician did because the chat logs haven't been released, but it's clear it wasn't something like >find a counterexample now, no partial results, no wasted tokens!!
>>539602100>This. Also math "problems" often remain unsolved for decades not because it's unsolvable by humans but because no researchers are really interested in the problems. People only work on what they're actually interested in. Not what is listed as a "millennium prize" problem or what's on the list of unsolved problems.That too. These articles (and faggots/bots like OP) are meant to spread the illusion that AI is already one-prompt solving major problems in Ph.D.-level mathematics, but that's not the case. And the lie is made all the more credible if STEM celebs like Carmack and Tao are paid to grift it.
>>539603200>>AIThere is no AI since there is no intelligence to be found. It's an LARGE LANGUAGE MODEL. It's a statistical model which is based on the frequency of words and their order. It works because human language is highly predictable. When you know the first few words it is easy to determine what the next words can be. And then it's constrained by other patterns to get outputs that are consistent with the prompt it is generating text based upon. If it is running correctly, the output will always have the same meaning in human language but with grammatically correct variations.
>>539602100>The only new thing is statistical model used to generated statements.and decisions based on those statements. those decisions are used for tool selection, where, for example, one tool is a web search and another tool is calling sympy (python computer algebra package)
>>539593640what?these are literally circles with numbers and letters with some arrows drawnit means absolutely nothing
>>539595904Yeah I find it most useful in terms of revealing those additional hidden details, giving them form in language, whereas before I may have only been internally aware of them, now they are better presented
>>539593640>some 1990s graph theory conjectureYeah... it may solve some graph theory or algebra thing here and there (as another anon pointed out we still have to see the chat logs), but generally it sucks at even graduate stuff. And what is ChatGPT's score on every year's Olympiads (in which some random Asian kids get great scores)? Is it still 0?
