AI is replacing mathematicians.
Nigga we aren't going away. We will still be digesting abstract nonsense even if we don't need to produce it by ourselves.
>>17039512whos gonna pay you my dude? ai?
>>17039550>he thinks all mathtards are employed lmao the joke existed long before you. But no those in academia will still be getting paid. Outside they always had to find some other job to feed their mouth.
>>17039485This board was actually somewhat decent before these AI cultist midwits infested the place. Unfortunate.
>>17039485we needed AI to solve this? looks plainly obvious to me
>>17039550>bending the knee to gov grant instead of decentralized off grid living with AI waifuTerminal NPC mindset, grim.
>>17039485Oh nooo, another graph slop has fallen?
>>17039485nooooo not the mathematicians nooooo!!! who is going to employ them now ??hahahahahahahahahahaha
>>17039641>Oh nooo, another graph slop has fallen?this desu
>>17039512This picture isn't a pushback argument.Math at uni needs funding.Math at uni needs people wanting to go into math majors.Yes, without mathematicians there won't be people who ask pure math question. And that's bad for culture. It's bad. But not an argument for it not going to happen.The contribution part will always be called to question now, too.The whore called Tao is the peak admiration a human will ever get from math now. It's been 4 months and the potential status one can get by proving theorems has dropped like a stone.With AI (and yadda yadda, it's as worse now that it's ever going to be), the second person interest in math skills will plummet, and the interest of young people going into it goes down notably as well.
>>17040241>The contribution part will always be called to question now, too.Computers have been used to solve math problems for quite a while, especially for finding counterexamples.AI is a cool tool, but it won’t solve anything without guidance, and it’s mostly useful for finding counterexamples.The hardest math problems remain unresolved, most likely because they require conceptually new mathematics.
>>17040241It doesn't matter if it turns into something like history or theology the math community must be sustained. Otherwise some obscure subfield is going to become extinct.
>>17040270I think you're underestimating where things are going.But even if there's no parabolic improvement, math also lives in part from math people doing math for others, and they will exceedingly not need that.
>>17040241>Math at uni needs funding.>t. yip yip i'm a good hostage puppy>t. rub mah bellyWe're sick of being told to fund heckin' next newton and then you turn around and promote Jason Arday instead, and sign letters to fire people like Cofnas who actually care.You don't care either, coward faggot. Fuck off.
AI replacing everyone, including mathematicians, kek
>>17039485raping*
>>17039485>we spent billions of dollars training a machine to brute force something we coulda paid a grad student 2 cans of tuna to doAnd their only proof that an AI did it is a summary of the AI's thoughts written by another LLM
>>17039485Okay, try doing some real math. There are 2^273 binary colorings of a 3x7x13 matrix: how many of those have no monochromatic row, column, or hypercolumn?
>>17041373How about no, and go suck my digital balls. Feed more money into my data centers, peasants.
>>17041405So far, AI is batting 0000 on any real math.
>>17039485Is this really something a brute force algo could not have found, or is it just something nobody bothered to?>>17039550Who pays gender studies, you absolute clown. Mathematicians will never be irrelevant enough for most first world governments not to pay them. The US, who is the only one pushing AI slop because they pump and dump their failing service economy with it, is the only one insane enough to try and slash their positions, which would be an even clearer signal for the rest of the world to stop sending their qualified people there.Most of you, even many qualified scientists, don't see this sad affair for what it is, the swan song of the US hegemony losing their academic and economical lead.
>>17041713It is brute forcing a semantic space. It lacks a guarantee of working eventually. The semantic space could simply not be rich enough. But it is also too large to ever know for sure. You get some m limit computational expense, like digging for gold to strike rich. If it takes more than that much, then move on to a different problem.So this is brute forcing the low hanging fruit as a marketing gimmick.
>>17042041I mean specifically conjectures in graph theory like this. It's 8 nodes with a handful of connections and small numbers. I understand how unreasonable the combinatorics are getting but this looks like something a heuristic algorithm should have found 10 years ago
>>17042041Brute force is trying every combination. Not possible in this context of infinite space. You don't even know what brute force is and computation isn't new. Beside, brute force itself is a valid way to find solutions, but this is not the case
Usecase for Dinitz-Garg-Goemans conjecture?
>>17042041>brute-forcing a semantic spacealso called "thinking"
>>17042169Okay, try "thinking" about some real math. There are 2^357 binary colorings of a 3x7x17 matrix: how many of those have no monochromatic row, column, or hypercolumn?
>>17042189If an LLM can not do it, it is not valuable maths. Very simple
>>17042041>brute forceBrute force would imply it's trying to check everything iteratively. It's more of a heuristic search algorithm searching through a huge library of highly organised data. Think of current AI as the stereotypical asian kid who memories math. Simply what current frontier models are doing is trying to come up with a proof that it can glue together from existing math.
>>17042189Eat shit:>>17041515
>>17042055You are a retard. >>17042169Can you show me the thinking flag in the control word of any graphics card plox? >>17042283You are correct that their approach could be and likely is looping and inefficient in some regards. But that doesnt really matter when their goal is headlines and banker grifting.
>>17042403No one cares about an estimate; exact number or suck it.
>>17042413Alright, it went and hacked LineShine to run its script, now there are suited Chinamen knocking very insistently on my front door, but the exact number is 1.856278...360042 x 10^111
>>17042481
>>17042484I accept your concession.
>>17042486>Indian debate syntax
>>17042487The only funny thing about your spamming is that next gen's LLMs will be trained to regard binary colorings on a 3D matrix as an important problem type.
>>17042490That's definitely not the only funny thing about me spamming an unsolvable, yet solved, problem to the AI worshippers.
>>17042189Total possible colorings (2^357):293567822846729153486185074598667128421960318613539983838411371441526128139326055432962374798096087878991872Valid colorings (no monochromatic row/col/hypercolumn):184768612977464560854764345398638779283731088830870097458870629968133602274559121450655660710Digits in valid count: 93 decimal digitsFraction of valid colorings: 6.2938986700e-16Computation time: 37.89 seconds
ohhhh it's THAT guy spamming the matrix color stuff, i should've known
>>1704137335869368000284964747947883965910010334139032682291146917853754997486470 (71 digits)>>17042189184768612977464560854764345398638779283731088830870097458870629968133602274559121450655660710 (93 digits)
>>17039641it's a data problem, just wait until they start walking around and/or corporate employees are instrumented and they develop a geometric intuition.Give it three years.
>>17042529These are both wrong, fyi
>>17042809nope
>>17042814yep. SorryIt will amuse me to no end, though, if AI jeets simply build an entirely fake world around an old, unpublished result that AI will never be able to explain.
>>17042819You can't bullshit me, the results pass all sanity checks and are the same from 3 different solver scripts. At this point I rather doubt that "old, unpublished result" you rely on.
>>17042823>At this point I rather doubt Of course you do; nevertheless, the numbers you posted for 273 and 357 aren’t even within a mile of being correct. Lol.
>>17042826No, you're just bullshiting. The results are in line with the expected estimate. The solvers blindly reproduce your result for >>17042232 exactly, and I assume every other result you didn't declare wrong until now was correct, so there's no chance of an error. There's simply no reason to believe the results are false. This is a desperate last attempt at trolling because you're malding about your shitty problems being solved by a free chatbot which, you claimed, shouldn't be happening.Actual researchers trying open problems use models/setups that are effectively an order of magnitude more capable by the way, even if this 90 IQ bot somehow missed a critical bug (it didn't) a pro model would solve it, I just don't want to spend $30 bucks to shut you up.
>>17042841Nope. Still a complete failure. Lol.
>>17042844This is incredibly pathetic, but then so was spamming /sci/ with your pet problems so I don't know why I expected anything better from a troll.
>>17042845>This is incredibly patheticYes, you are. Sorry. So how about you output the easiest example, 105, which most people can probably check themselves. Rofl.
>>17042848There's absolutely no point since you're not interested in truth, but it doesn't cost me anything:39581199709775325543313239 (27)
>>17042852Kek. Your output is an odd number.
>>17042854Yes, it's missing a 0 at the end, dipshit.395811997097753255433132390
>>17042855https://warosu.org/sci/thread/15735816#p15735827I’m potentially impressed. Do 3x5x11, which isn’t searchable, and I’ll undo all of my hate and proclaim that AI can do real math.
>>17042863877199530806922247135399990472579925903590 (42 digits)
>>17042865You win. I lose. AI is top dog
>>17042871Well thank you. To be fair, it did take 3 tries for the clanker to produce a decent solver, the first one was too simple and overfitted for 3x matrices, the second one was good enough for 5x but still took minutes for 7x, and only the third managed to solve those in 0.x seconds.
>>17042875Can AI actually output a formula for 3x7xX? That would be cool.
>>17042877Assuming 4chan can render it properly, yeah[eqn]N(3, 7, a) = \sum_{\mathbf{n} \in \mathcal{P}_7} \binom{7}{\mathbf{n}} (-1)^{F(\mathbf{n})} \cdot \Big( \Lambda(\mathbf{n}) \Big)^a = \sum_{\lambda \in \text{Spec}(3, 7)} c_\lambda \cdot \lambda^a[/eqn]for the actual solving it variously used slice PIE, S_A x Z_2 symmetry orbit aggreg, dual-axis multiset evaluation and vector condensation
>>17042881>SpecLol, but you already won, so that’s fine, I guess
>>17039485No it's not. It is as retarded as the people who blindly trust it.
>>17043677How do I tell if I'm retarded, help.
>>17039485Yep, it's over:https://arxiv.org/html/2608.02859v1>First, we saw AI models master textbook problems. Then they began to find counterexamples to longstanding conjectures [14, 20]. These advances, remarkable as they were, left room for mathematicians to do the hard work of writing and explaining novel ideas. Entire papers are a different matter. Corporate AI models now exist that compose articles faster than humans can. Their owners at Math Inc., Google, and OpenAI have openly declared their intent to make mathematics the proving ground for their envisioned future of automated labor [11, 12, 14]. This week alone, OpenAI posted solutions to ten human-level problems [15]. When these models are made widely available, our profession as we know it will be over. We are not prepared.>Tsimerman takes the view that AI improves upon human mathematics to its logical endpoint. In the last two years, he only took graduate students interested in AI; then, he no longer took graduate students; then, he left number theory to work on AI research [8].
>>17044536>Even before AI entered the conversation, mathematical knowledge was being produced at a faster rate than it could be profitably absorbed. Many good papers have never been cited, and journal editors struggle to find even a single referee, even for great work. This problem is about to get much worse. Autonomously produced papers will break the journal system.>Tao has argued on social media that “some of the prestige previously awarded to being the first to generate a proof will need to be transferred instead to the humans who successfully verify and digest such proofs” [19]. But checking a flood of AI-generated papers for correctness doesn’t sound prestigious. It sounds boring. AI can generate artificial papers by the thousands. How many of them would you agree to referee? It seems more likely that corporations will offer the same solution as they did for proof generation: refereeing duties can also be taken over by machines. It’s only a small step from the increasingly common practice of using AI to scan a manuscript for mistakes.>The point is illustrated by recent events surrounding the Jacobian Conjecture. [...] Tao posted a “digestion” of the new counterexample on his blog, clearly meant to illustrate how mathematicians can be of value in a world of automated research [20]. But Tao discloses that his “digestion” is itself a report of AI-generated thoughts that he obtained in the writing of the post.>In chess, it is well-known that any novice can beat a grandmaster. The trick is to play against two grandmasters in alternating turns, playing their exact moves against each other in order so that the novice is solely an intermediary for information. The novice is thus guaranteed one win and one loss (or two stalemates), but no one would say that their win is meaningful. Without intervention, the AI-written, AI-digested future of mathematics relegates humans to the role of messenger.
>>17044538>>17044536More blatant fearmongering which can be safely ignored. At this point I genuinely don't give a shit and no one else should either.
>>17044538>journal editors struggle to find even a single referee, even for great work.lol. Someone is sneeding hard that their multidimensional hypergenders paper wasn't recited in Logos magazine. What type of smug journalist would even put out such a stupid line. >some shit i dont give a shit about actually matters a whole bunch>and you can tell because nobody talks about it>and I wont give you any specific example of it>AI bad btwThe leftist is incapable of argument.
>>17044566I guess skew pentagram maps are satanic and not straight
>just fuck and let Tao became bento deliveryman>it run through wang hong's paper so have sex with herthat's why ai rocks, it generates my fake data and paper on google with corpo drone trust
>>17043679Around 90-95% of complex math done by ai is comically wrong It trends towards midwitism Like it will talk about shit like the universe being 9d crystals and shit. Imagine getting most of your math knowledge from people from India and china lol. Yeah it’s badIndians not in India are great but the country itself is mostly full of pseudo scientists china is much worst
>>17044699>ESL obsessed with indians and chinkshmmmm...
>>17039629Your erotic ai ap is not real
>>17044699>Indians not in India are great
>>17039485>LLMs everywhere, in everything, still no profits>"rapidly growing market, invest now! You're missing out! Once AI ...">google's quarterly report came out, negative profits from AI overspending>"that means nothing luddite, learn economics. Once AI...">look at "vibemathed" that was constantly brought up >AI "discoveries" have been flatlining for weeks>"stop noticing things you stupid leftist and buy my bags! THIS IS NOT A BUBBLE, ONCE AI .."uh, sure thing bud, two more years I'm sure.
Meanwhile the CEO of math.It's going to be like speed running. Good speedrunners can make money on twitch and youtube.Beating a game fast without tool assist is always going to look impressive.
>>17046665He has a point, a good point even. But people don't learn without consequences, they just don't. The community will need to suffer and then correctly pin the blame on the locusts in their own ranks.
>>17046665Tao having regrets? Holy shit
>>17039485Mental illness, """social media""" and corporate marketing are replacing your connection to reality.
>>17046665Yeah how are they going to enforce it? Lobotomise models into not answering certain problems? That sounds so stupid and pointless also anti intellectual. Though I do agree if we don't change academia properly zoomer mathematicians would be more retarded.
>>17046880>stupid and pointless also anti intellectual.That's you and the entire AI cult.>Yeah how are they going to enforce it?By scrutinizing all significant results and treating offenders like the lowliest of academic plagiarists. If your solution to an unsolved problem just so happens to match the one produced by a slop generator, you're out. Simple.
>>17046880If it's not done by human it does not enter into pantheon of human science.Verifying will be like it is done with speedruns. You upload video of you yourself doing the proof from start to finish.Then people watch it frame by frame.
>>17046898>Verifying will be like it is done with speedruns. You upload video of you yourself doing the proof from start to finish.Holy mother of all brainrotted college kids...
>>17046890>>17046898this is hilarious, keep going
>>17046908I don't need to keep going. Serious mathematicians need to keep going (which they will), until it becomes academically unacceptable to submit AI slop results, at which point you won't see any more high-profile names posting on xitter about how they used ShabosLLM to solve an open problem and you'll lose your shilljob bumping worthless threads about it.
>>17046915>it becomes academically unacceptable to submit AI slop results, at which point you won't see any more high-profile names posting on xitter about how they used ShabosLLM to solve an open problem and you'll lose your shilljobthis. the solution, as usual, is to remove the personal incentives. it's easy in this case because you only need to disincentivize serious career mathematicians. xxAnonxx420 can claim to have solved the Riemann hypothesis using ShatGPT N+1 on social media but he'd be simply pissing in a sea of piss. slopgen marketers will also move on to shitting up other fields if they can't get any famous mathematician onboard with their claims
With AI garbage you can prove anything you want. >AI proves 1 is 2.Cool, but I'm not reading this garbage. Preprint can handle it.
https://www.anthropic.com/research/formalizing-fermats-last-theorem
>>17046940What’s this?
>>17046944It's an informal proof that "AI" """researchers""" are clinical imbeciles. For example:>Claude completed the proof in 11 days, producing computer-verifiable proofs of 30,300 theorems along the way (using 29,500 in the final proof). Dozens of Claude agents collaborated to define concepts, prove intermediate theorems, and use those theorems to prove ever harder statements. At 13 million lines of Lean code, Claude’s proof is over 5x the size of Mathlib, the principal community library of mathematical proofs this theorem builds on.So now instead of humans having to verifying a human-readable proof, they have to verify 13 million lines of AI slop. What's next? Is AI going to spare humanity this effort by generating 130 billion lines of Lean to prove that the Lean proof actually proves what the AI claims it does? Whoever wrote this (probably the same thing that wrote the "proof") is an obvious mental inadequate.
>>17046948what? aren't the only things you need to manually verify the axioms used and the final statement, and then if the program compiles, it means it's a correct proof, no?
>>17046948That's just formalisation of an existing proof. You do realise that vast majority of recent ai generated proofs are informal right? They are messy but still human readable. Eventually it will become much clearer. Lean proofs are just certificate that the proof is worth reading when in normal circumstances your paper would have went through peer review hell.
>>17046963No. You'd need to verify every single "concept" it defines to make sure they all mean what they normally do in mathematics. And if you think that means simply going over the code and comparing against your math textbook, you're wrong again. You can't just use generic human-friendly definitions to prove anything serious using a theorem prover. It all needs to be framed for a particular proof strategy, which itself needs to be framed in terms of how the theorem prover works. It's easy to distort definitions to suit an intended result and hard to prove that didn't happen.>>17046968>That's just formalisation of an existing proof.How do you know, retard? Did you go through 13 million lines of Lean?
>>17039485No math isn't dead, AI is just doing what everyone predicted. It's solving problems which are unreasonable for humans to solve. That doesn't mean it's not solvable it just means that you'd probably get fucking fired if you told anyone that's what you're primary work was.Oh sorry by everyone I mean everyone who isn't fucking stupid.To be honest this is way less concerning that the possibility of professors becoming replaced by AI. Because you could actually just use AI to create the best education for any human being with heuristically optimized pathing while probably producing students about as strong as top ivy leagues if not better considering how many drop out to do the way more intelligent thing (make a shit ton of money for boring work).Let me explain: basically there are only so many people who actually want to study math in college because most people fucking hate mathematics and avoid it like the plague. That's why I always say never trust a doctor who uses numbers over experience, which is to say doctors can't do math for shit. It kind of symbolizes the whole can't have your cake and eat too. So those undergrads are being piped through a very particular pipeline which DRAMATICALLY varies in quality like very very high variance in quality despite the fact that mathematics is largely the same forward and backwards. Which isn't necessarily bad but the way it's implemented in most cases destroys the value of variance since it largely just affects quality in terms of ability ot contribute meaningfully.In general an undergrad will probably never see graphs in a serious way. Maybe they will see it in completely pants on head CS way in a CS class. A masters student is just trying to go in and out. Meanwhile a PhD student is usually trying to suckle off an advisor and the advisor isn't working on that bullshit cause it's stinks. There's this say in finance: a dog with fleas. Only normies and doomers are shitting their pants cause they're retarded.
>>1704694029500 theorem were first formalised for the whole proof btw
>>17046991Peak slopo archetype They are farming the lowest of the low hanging fruit by drawing connections nobody cares about and will literally never read nor check. And it will come out when some sperger uses it to build a major proof and some schizo just doesnt believe it and deep dives the slophole.
It appears that navier-stokes has been solved
>>17047093well post a fucking link
>>17047094Elliot Glazer @ Epoch AI (OpenAI funded Epoch AI's 'Frontier Math')
>>17047138So why aren't they releasing it? Are they holding it for IPO day? Seems stupid when openai might punch them in the guts.
>>17047138How’s that even possible? This is not something you can brute force, so it’s probably nonsense.
>>17047093>It appears that navier-stokes has been solved>It appearsYou're seeing things. If "AI" is so smart why can't it solve the escalating AI psychosis problem?
>>17047138Oh ho
>>17039485Calculators have replaed mathematicians decades ago. They still have a job.
>>17047248A "calculator" used to be a person with a job.
>>17047300>a woman is a person@grok is this true?
i work with mathematics and I use AI. its a pretty good tool, but a human has to digest the slop with infinite mistakes and make good prompts. else its useless.
>>17047316>do a breakthrough >proofs that have never been read by humans but come with leanmath belongs to the clankers nowA human has no role in this process clankers will shit out solutions incomprehensible to humans and that's a feature not a bug - super intelligence will not be held back by meatbags throwing hissy fits about how it's important that shit be explained to them like they're children
>>17047320This is true
>>17047320>clankers>a feature not a bug>super intelligence>meatbags >hissy fitsMentally ill American downie detected. Why not rangebanned?
>>17047320it's fine, the clankers have infinite patience and will try to dumb it down for our dumb asses too (for a price)
>>17044699>Indians not in India are greatFuck offGypsy
Math was created by Gypsies btw
>>17047320This. Why can't you people understand you're no longer part of the process at all?
>>17047371The same reason your doctors can't understand your psychotic rambling and keep you in the padded cell when you get too mouthy.
>>17047320Oh yeah who is going to audit your definition and axioms. You're gonna ask the clanker to solve problem A and it made up B and you gonna claim it's A because you lack the basic understanding retard. Also what's point staying locked in your palace while you can't see your kingdom? Just gooning and eating isn't life there's joy in learning and understanding has always been fundamental to mathematics. If there's no interpretability then it's just no longer math even if the oracle able to pull off application. And most won't even have application so basically "If a tree falls in a forest and no one is around to hear it, does it make a sound?". In the end it's nothing.
>>17047398>your definition and axioms>You're gonna ask>your kingdomwhat part of 'A human has no role in this process' do you not understand?
>what part of 'A human has no role in this process' do you not understand?
>>17047406God damn illiterate AI bros are exactly why all of us are given bad name. Don't worry in the future you will be devolved into the monkie and return to no thinking zoo.
>>17047185He doesn't know
>>17047398You don't even know how electricity or LCD screens work but that doesn't stop you from enjoying life.
Hodge Conjecture has been solved by AI. It's now being reciewed by the top mathematicians in the field.
>>17047664No it’s not Cockglazer confirmided it’s fake
>>17047664We have our top men on it. Top. Men.
>>17047650So gooning and eating slop forever mutt bitch? Some of us are more sentient than you and want other things than just being a mindless goy.
>>17039485>AI is replacing mathematicians.fucking based
>>17047183Hi
>>17039512Now you are a button pusher.
>>17039613>This board was actually somewhat decentnope this is not true this board is completely worthless just like 4chan in general
>>17050228They can still swallow those slops while you can't
>>17040349You are such an imbecile, you and your family deserve to be killed and oh boy you don't expect how many people will celebrate when we start killing racist retards like you once society collapses. Cofnas was literally a dei hire, YOUR HERO IS A DEI HIRE. Arday received too much backlash from racist bad actors, white authors who have plagiarized their work or conducted faulty research even in medical fields leading to deaths don't receive as much backlash as Arday did. You are worthless
>>17039512Explain that to those who fund you. :)
>>17039485who did openai copy this one from?
>>17050520explain whos going to pay you when no human job left
>>17051368>he doesn't own SPCX and TSLAngmi
>>17039485It's not replacing them.It's augmenting their abilities.And unless somebody disproves Gödel, there will never be an "end of mathematics."
>>17039581COPE LMAO, you will be worth NOTHING relative to other people.my dude, literally anyone will be able to do mathematics if a multidimensional being who's trained on everything you say and how you think can simply show you visualizations of the problem. FOR STARTERs the cartesian product visualizing functions (that are primordial on LEAN) will literally WORK FOR ALL CLASSICAL MATHEMATICs, so what do you do? When everyone has an army of 3blue1brown trained on you, do you realize what this is going to be? AHHAHAHAHAHHAGENETIC INTELLIGENCE WILL BE EVERYTHING AHAHHAHAHANO ONE WILL BE SPECIAL MY DUDEyou don't realize how much this hurts your ego, all the people who are less intelligent genetically will simply be shown exactly why and no one will care, there won't be pride AT ALL
>>17052531That's good if you wish to learn however if you don't even bother doing that instead spend doomscrolling and playing vydia then what's point?
>>17039485Did anyone ever really care about this enough to work on it?
>>17039485>again a counter-exampleIt's all just brute force.
