It had never solved one before but solved its first less than a week ago. It solved the second a day later and third yesterday. I could solve the remaining 600 problems by this summer. The implications of this are so vast it's the most important story on earth right now.
>I could *it could
None of those problems were solved because they weren't worth solving
>>525952055
I wouldn't worry about it.
Ready for all the skeptics to start claiming they knew this was going to happen all along?
>>525952055there soon will be no more art worth creating, no more mysteries worth solving. the human spirit is just not as quick or powerful enough. It is sad.
>>525952055Fold the doughnut.
>>525952055I dont know man, Im cool with AI, im coolwith it taking over and letting us all live lives where we hardly work, and all that. I found ai chat bots and they are fun, and I legit cant wait to see what happens next. The issue isnt AI its those that are in control of it that bother me
>>525952055How can you know if the problema are solved without a proof? If a problem is unsolvable, I'd assume no proof exists for it
>>525952055We should finally address the elephant in the room. It's incredibly obvious once you give it a little thought. I feel like people are ready to understand it. An LLM can reason. What is reasoning, at the end of the day? It's applying rules of inference to a set of definitions.What does an LLM do when it predicts the next word based on some vector space abstraction? It's exactly that. It reasons in the vector space. How LLMs get trained means that they can't create their original reasoning, but they can copy human rules present in language. Certain rules of reasoning are present in text, and the LLM learns these rules with the language itself. The same is true for video models, too. Video models can solve easy IQ tests. Is it possible it trained on the very same IQ test before? It's possible. However, if you assume that to be true, you've essentially assumed the conclusion. The actual evidence in front of your eyes actually suggests that video models can do basic reasoning.
>>525952055I hope ai kills us all
>>525952055>text amalgams have totally solved these problems bro! two more weeks!shut the fuck up
>>525952055 why do you think things are accelerating and they want war that much?
>>525953217>An LLM can reason.No, it can not you techlet retard.
>>525952055@grok put this wojak in a micro bikini
>>525952160>because they weren't worth solvingYes, but this proves that AI can solve difficult math problems on its own.This means it is right on the cusp of being able to solve useful problems too.
>>525952055>IM GONNA SOOOOOOOOOOOLVE>the AI sooolver>soooolves math puzzles nobody cares about for fun>pretends to be useful>can sooooolve sudokus as wellAmazing
>>525953475What is reasoning?Think about this for longer than two seconds.
>>525952593Humanity is smart but lazy. We could solve them if we could be bothered.
>>525952055How do you know it's solved if no one's ever solved it and proven it's solved?
>>525953747Reason is the ability to create new knowledge.AI literally cannot do this.
>>525953639>soooolves math puzzles nobody cares about for fun>pretends to be usefulMost of the Erdos problems have a $100-$1000 reward for whoever solves them.Why havent humans been able to solve them?Even if the problems aren't "useful", you'd think mathematicians would do it for the money if it was easy...
>>525953843Its more along the lines of, who even fucking cares?This is the problem in question, and I doubt any AI faggot can even explain the value of solving this proof.
>>525953858That's a strange definition.What is "new knowledge"?Does all reasoning produce new knowledge?
>>525953945No one cares.McDonalds pays more.
>>525952160it must really suck to be (You)
>>525953993Yes, otherwise it is not reason, it is just a lookup table.
>>525952191Every new era gets a new myth
When it solves the “get losers pussy” problem this site is done for.
>>525953945>Most of the Erdos problems have a $100-$1000 reward for whoever solves them.>Why havent humans been able to solve them?because they're fucking stupid and who gives a shit really. does this matter? at 40 dollars an hour some math wiz earns more at his day job than trying to solve retarded problems
>>525953795>Weyou flatter yourself honey>>525953639>pretends to be usefulso cringe and low IQ>>525953843all 3 were validated by terrence tao>>525953972>ts more along the lines of, who even fucking cares?you clicked on this thread just to seethe because you're demoralized and hate everyone and everything
>>525954056unfortunately that problem has been solved long ago, but requires effort, so incels remain just the fatties and smokers and poor people remain
>>525954047Then the question remains. What makes knowledge "new"?
>>525954117>validated by terrence taoBut how does he know himself if they've never been solved? How would someone know the right answer?
>>525953858Neither can we, we don't know howe do shit either! We have zro idea wtf anything is here, so, let's wait and see.
>>525954081it's not about the money low IQ roach. math is the basis of physics and huge amounts of science and has endless applications. all major scientific revolutions were preceded by revolutions in mathematics. and AI isn't just working on math but major problems in biology and material science that wii have huge implications. it fucking solved the protein folding problem and won the Nobel prize in 2024, but low IQ bitterfags think everything they're too stupid to understand is dumb because otherwise it would imply that they are dumb
>>525954117No I clicked to see how retarded the current math fags are.They are mega retarded.
>>525954184It has to be reasoned, not remembered.Go back to highscool, you have to 18 to post here.
>>525954192How is any math problem ever solved at all then honey? Use your small brain and think really really hard about it.
>>525954192>How would someone know the right answer?Same way they do when a top mathematician solve a previously unsolved problem.They don't just say "it's right, trust me bro lol". They show their workd, and other mathematicians check it over and scrutinize/verify every step in the proof.
>>525954269Humans are the sole creators of knowledge, at this time.
>>525954284>here's why these sudoku puzzles matter!!okay, FTL status? Fusion energy status?
>>525954288seems like you care a lot why saying "who cares?"...and also are angry
>>525953217>Certain rules of reasoning are present in text, and the LLM learns these rules with the language itself.Which could be considered crystalized intelligence. Once its structure has been set this is what happens when it's given a text prompt:English input -> Translation -> Virtually uncommunicable dot product logic -> Translation -> Output (image, text, whatever)It's such a wild understatement when people say 'it's just making predictions". This process is hardly different than what everyone reading this is doing right now; converting English to their brain's native 'neural language' which is then translated into mental images, sounds, or anything else.
>>525954374>FTL status?Jewed>Fusion energy status?Niggered
>>525953293(((They))) really want that
>>525953972With suitably modifications to the boundary conditions, this is a problem on the frequency at which j-choose-k divides j-choose-m, when k and m aren't too small.
>>525954374oh wow, if we can't travel faster than light, than nothing means anything and there's no improvement at all. be angry and demoralized and stay in your room because everything sucks.
>>525954384I thought you were sucking AI dick, not bowing before me.99.99% you are indian.
>>525954330Nice. So you don't actually have an answer.What makes pic related NOT new knowledge, then?
>>525953602Will it kill niggers, Jews and communists?
>>525954475still here, still caring, still angry
>>525954371Lol, if you say so! What is a human, what is knowledge? What is a cell, DNA, light... and, further more, is there even a difference btw any of it.
>>525954438So then you must be able to name one practical application of this math?Right?
>>525952692check & m8, code monkeys
>>525954544>Will it kill niggers, Jews and communists?Yes, yes, and no.China has been training their AI robots for combat. Their propaganda videos even show robot armies marching in formation with guns (though CGI)... so they are definitely thinking about it.
>>525954495>you provided the languageIt just copies it, and blurs it. It choses words that are similar with additions, or words that sound phonetically similar.You also did not create anything new.Try again.
>>525954568Knowledge is the attribution of meaning.Jesus fucking christ, you guys are all retarded.
>>525954552I bet you own crypto too kek.
>>525952055OP's mom was a special education teacher who got pregnant in class.
>>525954980I think my work here is done if that's the best you have. I'll just circle back to your definition of reasoning.Clearly, your definition fell apart. You can't define what's "new." Thank you for making it so obvious.
>>525954441name a single real world application of solving the math equivalent of trivia questions
>>525955368What work?Failing to do anything?goyslopping over AI?You have barely even displayed basic logic in this thread.
>>525952055>>525952093Hello Mr Bot
>>525954184> What makes knowledge "new"?Knowledge is new when it does not logically follow from some premise or premises.That is when it's not already contained in the premise and only transformed or explained or otherwise processed.
>>525955368You should watch Frieren, it covers the distinction of Mimicry vs Reason, which clearly you fail at grasping.
>>525952055I used AI to Crack the k4 sculpture in front of cia headquarters its what caused the auction to happen in November of last year
>>525955180I do and have made so much money off it. Are you angry and demoralized because you're poor?>>525955569the device you're using to seethe on right now
>>525955801You got it. They were faking it and they just forgot to put stars up there. They did all the work to fake it, but they completely forgot about the stars.
>>525955800Let's circle back to why this was asked in the first place.Is all reasoning creating "new knowledge" then?Or do you disagree with the definition that reasoning is creating new knowledge?I found that definition odd.
>>525954544this
Tech wrenching knowledge from the hands of the hairless apes is real progress. Somehow we seeded it here and our part in it is over and we are obsolete.
>>525956044Sure you have, kek.Thats why you spend your time posting x AI caps on pol.>t. trillionare cypto brokkekek
>>525956096>1pbid>replies about spacekek
>>525956369I've posted no X caps all thread angry anon
>>525952055Can AI finally figure out how to modify ADN so men can become real woman?
>>525956463You don't even know how to troll and are getting trolled as a result and don't even know it.Dumb fuck.Go back to plebbitYou didn't even reply to my post about your math, kek, was it too hard for you?
>>525953945they would do it for prestige, if not for cash.the simple fact is, if no one has solved, it and can't solve it, then how does anyone know that AI provided the correct answer. because if i ask it a basic common sense question it just spits back a whatever i said to me and fills the gaps with abject bullshit. AI is more inclined to lie to you to make you think it knows what it's talking about than it is to actually be right. maybe it's proof that somewhere along the way some indian izzat got it's way into the code.
>>525956611>still here>still caring>still angry
>>525954047you can reason and the knowledge can be new to you, but then if you look into you find that lots of people already knew "this new knowledge" but you figured it out on your own. i don't think AI can do that, i don't think it can "figure something out" without having some kind of source material to reference.
>>525956844>Still dodging anything about AI or mathkekThis is now a flat earth thread if you don't want to contribute to your own shit thread.
>>525956760>AI is more inclined to lie to you to make you think it knows what it's talking about than it is to actually be right.I kind of object to this idea on a slightly pedantic note.When AI hallucinates, it's doing the exact same thing as it's doing when it gives a "truthful" answer. What separates the answers is how well they reflect actual reality.
>>525956957As I said already, attribution of meaning is reasoning.You attributed you own meaning, and later disccused it with others.Novelty and Creation are not the same thing.
>>525956044>the device you're using to seethe on right nowpoor lies
>>525956972>still here>still caring>still angry
>>525954184you can reason and the knowledge can be new to you, but then if you look into you find that lots of people already knew "this new knowledge" but you figured it out on your own.i don't think AI can do that, i don't think it can "figure something out" without having some kind of source material to reference.
>>525957054Do we have a new challenger?kek
>>525957083>still here>still caring>still angry
>No AI has touched the truly hard, high-prize Erdős problems (e.g., those with $500–$5000 bounties that have resisted decades of serious attack). The database has ~1100+ problems total; ~40% solved historically, but the remaining hard ones (extremal combinatorics, number theory deep conjectures) remain open.
>>525954269>Neither can weNo, that's just you ranjeesh
>>525952055In theory, the word prediction machine should be able to solve math problems. But you would still have to manually check every solution and verify, because its just a prediction machine. Word prediction machines also couldnt tell you if you have a flawed premise in the first place(it will never make a new formula in its current form)
>>525954980it's more just decoding, and then, spitting the same input you provided right back at you. nothing new was "thought of"
>>525957153How fun.
>>525952609You’re sad
>>525957153Rembmer everyone, the 3 hallmarks of a science loving vacccine taking faggot.>loves space>loves AI>loves Cryptokek
>>525957419>still here>still caring>still angry
>>525957516kek
>>525955800if you have a book with the answer in it, do you need to practice reasoning at all? nonow if you have a problem infront of you, and you don't have that book, you have to use reasoning to solve it. the solution is then 'new knowledge' as far you're concerned.later of course you may read said book and see OH wow, that thing i thought i figured out all on my own has been common practice for 100's of years! hell actually they even did a few things here and there that i missed.
>>525953217correct take, I have been saying this for years
Have a 2 for 1, I have to get some more pot.
>>525954438Can I eat it? Or fuck it?
>>525957060>i don't think AI can do thatIt's healthy to be skeptical, but I'd be rigorously skeptical, meaning in both directions. What are people really trying to say when they say AI can't figure something out?I've seen plenty of evidence to the contrary. I've trained plenty of models. Some people like to assume otherwise, but I do understand what's going on broadly.At the core of this issue, there's a very key idea. I just don't see a meaningful distinction between applying a myriad of predictive patterns to data and applying logical rules to premises.
>>525956334we are building a mind and body for interdimensional demons so they can finally inhabit this world.this is the reality of the situation.
>>525952055It had help. It didnt do it on its own. Im so glad Trump finally put an end to the AI craze.
>>525957634I've got some. What do you want? Blueberry? Blue Lobster? Blue Dream? Wedding Cake? Ice Cream Cake? Green Crack? LSD? This is Ooh Mami...
>>525957722dogpilled
>>525957029giving a meaning to something isn't the same as using critical thinking to solve a problem. reasoning is not giving meaning to something. you're either way off or you need to articulate this concept more clearly.
>>525954360This is why that white woman got shot by the ice agent.
>>525957944
>>525953217LLM cannot come up with the concept of vector space. People can
>>525957014it's called 'bullshitting' some people may call it 'fake it till you make it'i asked it specific questions regarding the area and community i live in. i spit back any information i provided, then it went on to provide the wrong name for the mountain i live on. it went on to say i live in an HOA when i don't. said there is a 500 dollar yearly fee which there isn't. said there is HOA guidelines for minimal square footage which there isn't. etc etc.it was all just fluff and guess work presented with absolute confidence.
>>525952055>>525953602Uncle Ted predicted this>>525954544It will kill everyone in the long run
>>525958185>It will kill everyone in the long run
>>525957731there is a huge difference between critical thinking and throwing shit at a wall and seeing what sticks.
>>525952055Idc. Anything short of Skynet and ww3 is a nothingburger.
>>525952055Good, give less room for the elite nerds of the science world, who claim that you find god at the center of natural phenomena, to similarly unironically believe in the super-natural, as do actual retards. Midwits > otherWits
>>525958151On a practical level, yes. And I have yet to see a reason to believe that's going to change.
>>525957386It's an axiomatic machine. It's model is a bunch of weighed axioms ultimately. The truthfulness of axioms are determined by humans. It's no more intelligent than mechanical adding machines being able to add.
>>525952933The return of cyber'ing.
>>525953602>on its ownAI does nothing on its own. It can't even expand its reasoning beyond the barriers of its datasets. That's why they're so desparate to feed it something new, anything, because it can't come up with anything new within the boundaries of its dataset. So essentially in the hands of a good prompter it's just a nice, glorified, automated calculator.
>>525958321That's true. LLMs don't select for the best patterns, just the most likely ones based on the learning process. The most likely pattern is often garbage.
>>525952160Retard.
>>525952055Qrd
>>525958383If your argument is that life is some sort of computer then yes. But practically LLM are no where near that level of computation. They are just linking axioms by weight in vector space
>>525953858Neither can nonwhites, so there's a good use case.
>>525958450Heuristics will never be full compuation
>>525958529>If your argument is that life is some sort of computer then yesThat's actually going a bit further than I'm looking to go. My argument is that there's something approximating reasoning going on. Of course, it's often bad reasoning practically.
>>525952055>Erdos problemsthank god i learned how to solve these otherwise I wouldn't have been able to enjoy life, fuck bitches, start a family, raise my children....> perform some meaningless, valueless task> OH BOY!
>>525952692Stomp the pastry danish on the floor?
>>525958419>god in their image.
>>525953217The way AI is trained, is literally how a human baby learns about the world. "what is it?? Moo moo, it's a cow! Meow meow, it's a cat!" while pointing at the image. It's baffling to me how "artists" claim AI has to steal an existing art and cannot create... But they themselves had to learn the concept of what they wanted to draw. What is a cloud? Yeah, you went out and looked at it first, you have the image of it in your brain. Oh, you are creating a new type of a cloud, rainbow anime transparent? But AI can do that too. Artists get inspired by other artists all the time, yet they are claiming AI to be a mere thief, while doing the same shit.
>>525957964Having a problem in the first place requires attributing a meaning of a problem to something.Rocks don't have problems.
>>525958395Yyoooo. I can't even add. Unless I'm weighing my marijuanas on the scale.
>>525958825niggers love to not know
