AI is already better than top mathematicians. What will we do when AI outperforms Shakespeare and Dante?
Yes, but only if you’re a soulless brown techjeet
>>25537141AI can never replace mathematicians like other professions, math guys are Gods, 160 average IQ
>>25537141Literature doesn't have "problems" to "solve" so it's a moot point.
no one reads so people will be happy to have an excuse to not do so. "AI is better so why would i ever bother reading moby dick", not that these bug vermin would read AI novels or plays either
>>25537188The "purpose" of literature is social engineering, and I believe AI will be very good at defecating literature highly effective for this.
>>25537141> AI is already better than top mathematicians.Or is it? When you look deeper into these stories there’s a human trying to brute force a solution and the actual contribution of AI is always in doubt, either having to be tardwrangled by a human or having stolen the actual ideas or results somewhere along the way. What we were promised is a reasoning machine that you can present a problem to and have it solve it with “intelligence”. None of that is happening. They also claim these things are getting superhuman intelligence every other week so their actual predictions (following the Kurtzweilian singularity religion) is that in a month or two we should have all the millennium problems solved and in 2 years it should be solving quantum gravity. Except by previous hype proclamations we should have been there already. They always default to “bro it just needs MORE TIME” bro, more COMPUTE! We’ll all see! And indeed, if they were actually right, we would, and there would be no debate about it.
>>25537141It would have to write something original to surpass them and it can't. Until unless it actually grows a soul by definition it just can't.
>>25537141>What will we do when AI outperforms Shakespeare and Dante?Probably never. The recent ai progress is based on a reinforcement learning with verifiable reward. There is no way to do automatic verification for writing quality.
>>25537316>2023: They say AI passed a middle school math test, but it didn't really! HA!>2024: They say AI passed the AP Calc exam, but it didn't really! HA!>2025: They say AI won an IMO gold medal, but it didn't really! HA!>2026: They say AI solved a Millennium Prize problem, but it didn't really! HA!At this point if you don't notice the trendline you have to be blind
>>25537141>AI is already better than top mathematiciansAI is just an extension of computing, why did computers not replace mathematicians anon? The only thing AI can ever replace are the things its proprietors and other (((interest groups))) want it to replace which means that they make money and you get raped. If you fell for the AI scam you are a gay retard, end of story.
>>25537631You are retarded >>25537331I hope this is true
>>25537141Not for nothin', but, uhhh, pressing x to doubt.https://www.forbes.com/sites/sandycarter/2026/09/10/did-openai-steal-a-1m-math-proof-four-questions-for-every-ceo/
>>25537674You are retarded
>>25537141higher mathematics is gay and that was solved by a team of boffins using AI
ai is already better than humans at chess. yet chess competitions thrive because people want to watch humans play chess, not ai. same with the arts.stem fields are probably the most in jeopardy because they're utilitarian in nature. but people don't want to listen to ai music or watch ai films or read ai books. they want to see another person's vision, not some algorithmic slop. so there will still be a demand for art.
>>25537350If you don't understand the difference between an AI regurgitating existing information and actively solving new problems you're too stupid for this conversation.
>>25537141AI is not better than human mathematicians. It is simply better at finding counterexamples because it is able to spawn subagents and work on 10,000 different approaches at once until one of them works. That's how many subagents OpenAI used to solve this problem.Imagine we had 10,000 mathematicians working on this problem, each assigned to working on different approaches in unison. They certainly would have found a counterexample.The problem is 10,000 mathematicians perhaps would have taken longer than AI did (88 hours), since they need breaks and rest, and definitely their salaries would have cost much more (the token cost at API price for what OpenAI did was only around $20 million).The power of AI over humans is its near-infinite extensibility & speed. Things which don't matter at all in literature. AI poses a threat to literature only because it can flood the zone with slop and demoralised readers and publishers won't bother risking their time and effort going through it.>>25537350The big developments were the introduction of chain of thought reasoning, RL training on reasoning problems with verifiable results, and tool use. Which were introduced in late-2024/2025. The improvement of the models from 2025-2026 has honestly been marginal.
Even if it were technically superior to Shakespeare AI could never outperform a bad artist because art is by definition human. Without the human element, there's no point in reading.
>>25537779yeah this is the interesting direction this conversation goes to. Are those monkey paintings art? if not is it because they are not human or because they lack intent/intelligence? if the former, could an alien intelligent species not make art? if the latter, would an accident that creates something beautiful or meaningful not be art no matter what?
>>25537141india and israel ban when?
>>25537884god i wish
>>25537141How do you use AI for lit? I tend to get weak recs. Could I upload things from work, and have it rewrote like Dickens or something?
>>25537779the human element is humans making the selection
>>25537770>they certainly would have found a counterexampleand yet they didn't.
>>25537989Because they werent working on it because nobody was paying them to
>>25537720You are double triple retarded
>>25538016You are retarded
>>25537141Shakespeare stole so many of his plots I bet he would've used AI
>>25538961He also used collaborators like John Fletcher
>>25537141can a clanker write this
>>25537978"The critic is the real artist" - Piero Scaruffi
>>25539006he doesn't listen to much art so how would he know
>>25537350Kill yourself idiot
>>25537141
>>25537755Artist won’t be patronized outside the ruling class of bankers and technocrats. Those elites who are commissioned will make abominations only to tastelessly decorate the elite’s estates. Mass produced “art” will be made by those same elite families, with the assistance of AI, to keep the goyim’s socially engineered programming up to date.
>AI>Math>Shakespeare and DanteWorthless surrogate activities.
>>25537141The only way it fully replaces mathematicians is if it solves the P=NP problem in the affirmative. That would make mathematicians obsolete and we'd have whatever technology is possible within the limits of physics available to us as quickly as AI could build it.
>>25539539> oh look its another machine cultisthow about a lovecraftian response> what if the machine is a reaper and it turns you into a husk for commander shepard to gank
>>25537254AI would probably find way more effective and optimal ways to social engineer than writing shitty novels
>>25537254That's the purpose of Chinese literature, thoughbeit.
>>25539772You didn't understand the post to which you replied. Read more, schizo.
>>25537316you think math is so untouchable then how do you debunk Universal Approximation Theory?
AI writing is dog ass
>>25539998AI writing is currently dog ass,it has the rest of time to get better
prosefags actually believe this lol. that the value of a wor comes from its form ergo AI that can make something of equal form can make something of equal value
WERE IS THE MODERATOR?This thread is NOT /lit/ related. Like starting a thread with "What if someone wrote a book about my ham sandwich?" [pictured]Not related."What? People can write about whatever they want"We have the off topic meta thread for this crap
>>25537755AI isn't actually good at chess. Chess engines don't work like AI.
>>25540355Post hand
You guys really think you're gonna beat the text-producing machine at its own game? No. There is no way "out." Only through
>>25537350Pic happened at the end of 2025. How does it fit into your trend
>>25537141AI's single best use case is literally gooning. Great job guys.
>>25540361Cry moar you weak janny bitchPeople can write about whatever they wantGo fuck yourself you get paid nothing for your hall monitor duties.
>>25537840This is just pointless masturbation about semantics. What actually matters is that people don't perceive media generated by an LLM or diffusion model as self expression, so it's boring. I don't think the problem is necessarily that a machine produced it, but the type of machine that did. If we figured out continous learning and had AI agents that accumulated individual memories and skills over the span of years, I could see people find interest in that, as there would be a self to express. Media produced by animals or accidents is often interesting, but for different reasons from human media
>>25537141The current limitation of LLM writing is that it basically minmaxxes rhetorical & stylistic techniques. They have learned from inhaling English writing style guides that things like "it's not just X, it's Y. And that's powerful." are incredibly impactful techniques when applied appropriately and in the right moment, but LLMs currently do not have the discretion to know when to use these flourishes. All the hallmarks of AI writing are from a technical POV good writing, and it's exactly their overuse/ubiquity that automatically outs a written piece as AI-authored. Authors know when to save a good turn of phrase or grammatical inversion for effect at precisely the moment they want to emphasize something, but LLMs just shove these into every phrase they author.
>>25537141>AI artI don't think this is possible at this current juncture. I'm not one that believes in the human soul or anything, but I don't think AI is capable of two specific things:>ExperienceArt is, in my opinion, an expression of ideas meshed through, and synthesized in light of, your individual experiences. You naturally form a loose narrative as your life progresses that informs your many thoughts, and through this you can be creative. LLMs are just trained on masses of data with nothing of the aforementioned sort. It's not just a chronology of data, it's events that happen to you as a being, and decisions you have to make. You contemplate, regret, or reminisce. Until we can simulate entire complex lifetimes, I don't think this quality will shine through in AI-generated media.>Learned craftPart of what I appreciate in art is wonder at how an artist even came to make something, or how they learned and honed their techniques. If you have been to a gallery, you have probably experienced this feeling. A technical mastery is not only rare, but incredibly admirable. Who fucking cares if you train an AI enough to create something that can be judged as "novel"? It's all brute force.It may be possible in the future, but to be honest about my bias, I'm not inclined to give a fuck about it when that time comes.
>>25540021It 'gets better' by assimilating mountains of human work product fed to it by other humans. Oh gee, the most sophisticated computer program in history can process input and produce output? Whoda thought.
>>25537350/thread
>>25541586ask chatgpt to make this image today and it can
>>25538961This. By combining preexisting popular plots and themes he was essentially the AI of his time
>>25541967>It 'gets better' by assimilating mountains of human work product fed to it by other humansYes? This is how it works, anon>I'm a writer and I get better at writing by going to the library and reading a lot>FRAUD! If you were a REAL writer you wouldn't have to assimilate mountains of prior work! FRAUD! FRAUD!See how foolish this sounds?
>>25542028Because that image was already made.
>>25541931>I'm not one that believes in the human soul or anythingYou are a fool, then.
>>25541967Are you pretending to be retarded? That's what people do to learn too.
>>25542059You're only setting yourself up for disappointment or extreme cognitive dissonance when machine capabilities are coming closer to eclipsing those of humans. You are the fool for believing there is anything inherently unique to humanity.
>>25542063>You're only setting yourself up for disappointment or extreme cognitive dissonance when machine capabilities are coming closer to eclipsing those of humans. It will never be alive or conscious, so I don't care.
>>25542065Never is a strong word, my already coping friend.
>>25542069Guess your calculator is conscious too, right retard?
>>25542072Good argument bro. I hope that you know that my first post is confidently anti-AI.
>>25542074My point still stands. How do you determine an AI is conscious in a way you couldn't already do for a calculator? The only separation is complexity.
>>25542078I wouldn't argue that AI is currently capable of consciousness. The point of my first post is that if you can, in the future, simulate such a thing, alongside a lived experience, then it would be capable of creating art. My stance is that there is nothing unique to human nature that prevents something else from eventually approximating what we are.>determine consciousnessI promise I'm not being disingenuous: consciousness in my view (I'm no psychologist or philosopher) is just a collection of apparent mental faculties driving a being with stakes. Among the ones we value in humans above animals are some sort of self-reflection and metacognition. Calculators cannot do this, but instances of AI can perform some version of these things, even continuously on their own, but only as a task-machine. It is not self-aware, doesn't possess individual identity, nor does it exist in and feel the physical world. Instead, it merely takes in countless facts about the world and outputs what it is trained to predict based upon patterns filtered through a neural network. AI in its current incarnation and use cannot be conscious, because we have our own (unfalsifiable) inner-worlds and it does not. If we strove to approach that goal, for whatever horrible reason you might imagine, I think we could get very close. But it would not be through crudely putting the next version of chatgpt into a walking robot, but rather some other rapid technological advancement prompted by >le AGI.
>>25542168get stabbed and raped
>>25542198You're brown
>>25542028Post your results then
>>25542168>consciousness in my view (I'm no psychologist or philosopher) is just a collection of apparent mental faculties driving a being with stakes.Really dumb definition. If consciousness is anything it's unitary experience, first personness. At this level of examination Occam's Razor and appeals to objectivity are extremely weak. "The moon must be moist because it's associated with the tides" tier reasoning. >AI in its current incarnation and use cannot be conscious, because we have our own (unfalsifiable) inner-worlds and it does not. I lean towards this being true but we really don't have evidence either way.
>What if AI beats humans at writing?>But AI won't beat humans at writing
>>25541604>weak janny bitchWeak insult.Clearly I am not the janny, and clearly the janny doesn't give a fuck what this board is for, so off topic fags like you shit it up. Brain-dead.
>>25537141Didn't they get dued by the mathematician who actually solved it but used their LLM to proof read it?
>>25542213>gets impaled and sodomized
>>25542065neither will viruses
>>25544162Do you suppose there is an upper limit on how complex viruses can get? Biologically, physically? I wonder about things like that
>>25542168you're ignoring the chinese room in the room. It's ironic you say believing in a soul will lead to future cognitive dissonance when your stance will do that. You're right that consciousness is unverifiable, therefore as soon as gpt50 becomes apt enough to mimic a real consciousness with a real internal world you'll be among the fools who'll believe a collection of parameters has "reason". Your dissonance will kick in when you realize that as the computer is deterministic, and it has consciousness, then so human consciousness must be deterministic. There goes your reason, intelligence, agency out the window.
>>25537141>outperforms Shakespeare and DanteA trivial feat, I assure you.
>>25537350The thing about AI is it's a very sophisticated use of Linear Algebra to reverse engineer solutions by throwing millions of floating variables and letting a computer just brute force the most accurate calibration.Unless I'm totally misinformed about the field, and it's been nearly a decade since I was involved in it, but you cant simply solve unsolved problems through reverse engineering. Maybe you can form novel connections, new applications of previous methods, etc. but you cannot solve for the unknown because having the right answer is a prerequisite for training the data set.
Did anyone mention that the calculation cost tens of millions in electricity yet?
>>25543077No. It's very convoluted. Basically, a mathematician at Anthropic or whatever was working on the same problem and is claiming OpenAI used his work to find the answer. Someone who worked with him at Anthropic or whatever was part of the team working on the problem at OpenAI so he's claiming his methodology was copied. Also, they arrived at the same result the same day but OpenAI announced first, after Anthropic had already been hinting they'd solved one or more of the Millenium problems, so it's basically a nerd cluster fuck. Anthropic figured out how to get to the answer but OpenAI figured out how to implement it faster (but they specifically raced to beat Anthropic who were stupidly dropping hints and hyping before they had the result).
>>25537141I miss you Golden Gate Claude (a few years ago, anthropic fucked with their model to get it to talk about the golden gate bridge)
