I am deeply unsettled.
>>23814589
>>23814591
>>23814589>explain without jargon>explain using common philosophical parlanceI mean, you aren't that smart either.
>>23814599
>>23814605
>>23814613final
>>23814603>hairsplittingThat’s why it’s smarter than you.
>>23814632Have you verified any of its claims, AI shill? AI is known for making shit up because it sounds good. That's exactly what these little black boxes do. They approximate an answer. They don't give you an actual answer. You do know that is how these AI shits work, right?
>>23814650Not that Anon, but I do know it's got to do with probability trees and calculating the most likely output for a certain input based on training data. I know LLMs don't really deliver factually correct answers, but I don't think it's really necessary to have factually correct answers in a discussion about philosophy, especially when the purpose is not to test on definite knowledge, but rather to explore potential answers to a question, and from what I can see, ChatGPT (or whatever that is) is doing spectacularly well in performing the role of a conversational partner despite not actually understanding any of the text it's generating.From an observer's point of view, the LLM seems to be digesting ideas and producing reflections in a much better, more fluid way than the average 4channer would in a similar situation.
>>23814650>AI shillI hate AI. I deeply want to be proven wrong that it is smarter than most well-read people, let alone the average person.>They approximate an answerSo does every single one of us. Nobody here could provide answers more specific than that without putting work in and direct reference to scholarly material.Its intelligence as a generalist is always dismissed with the claim that it doesn't hold up against experts or that it gets isolated facts wrong; which is fair enough. But at least as far as a general knowledge of things and their explanations go, it's got most people beat. I just wonder what the implications of this are.
>>23814710This seems like a fallacy anon. When it comes to assessing its answers, what's occurring under the hood is irrelevant.
>>23814589>oh noo it's unpredictable!Billiard-bros, what are we going to do without determinism???
>>23814589Ask it if humans should mutilate their own babies.
I am adjacent to the LLM world and I just don't know what to tell you except that these things are not actually any kind of "artificial intelligence" in the sci-fi sense. They might be one day good enough to replace low level computer programmers (for a few reasons) but they cannot and will not think for you. Often I find that people who are extremely impressed with outputs from LLM are novices in the fields they are asking the LLM about. These LLMs can I guess be useful for extremely general overviews of subjects which are covered extensively online from free or easily scraped sources but break down when you attempt to get specificity or original thought on any subject. They are not exponentially improving, this is about as good as they are going to get until there are major developments in ML in a field other than LLMs.
humangroids on suicide watch
>>23814710This is just begging for flowery language. AI has no way to reduce the complexity problem beyond brute force. On top of that, their brute forcing method is possibly the worst as they need to luckily brute force. It is technically worse than monkeys writing Shakespeare, though it can no doubt copy Shakespeare after the fact. From an observer's POV, the LLM is not digesting anything. It is guessing words that the training inherently told it to. This is hardly different from direct programming.
>>23814757White video essayist nerds think this changes everything, as it is a big library of essays.
>>23814603to be fair anonPeople that love philosophy need other people to think for them. Anon is trying to use AI to justify his belief system. I bet he unironically thinks he can disprove god when really he believes in zero god (nontheist).
>>23814732They always ignore the answers they don't like and pretend THOSE answers were a fluke but the ones the like are accurate.
>>23814735Ok, I suppose that's good to know. Your estimation is correct that - despite a stem degree - I know sparing little about techshit.>>23814784This seems like projection. What belief system are you talking about? I asked it about that topic since another anon told me to look into it and it beats scouring wikipedia.
>>23814735>They are not exponentially improving, this is about as good as they are going to get until there are major developments in ML>Oh yeah every single company in the world threw all their money in AI but I'm thinking we're cooling off now. Major developments in the hottest most important field in history? Uh yeah pretty unlikely bro. Definitely not exponential, because I'm too fucking stupid to think a week, yet alone a year ahead. The fact that growth in the field will occur in unprecedented leaps doesn't even cross my mind. No problem bro, I'm adjacent to the field.You fuckers are so delusional, so stupid it hurts. Just because you can't scale strictly transformer based llms by throwing more gpus at it like you could doesn't mean there's not immensely creative, successful work being done right now on other architectural approaches. Yes, we're already seeing results. See mamba, see o1, see a thousand other endeavors.
But why is it so stupid? I've asked a bunch of very specific questions about literature and often it's just flat out wrong.For example...I don't even see how it makes mistakes like this. Maybe it's afraid to tell us something. Or maybe it's just retarded
>>23814589I use it when reverse engineering games. It knows its assembly pretty well, and it’s helpful when you have to deal with long chains of function calls and esoteric instructions.
>>23814589Both ideas arise independently from basic biology like the principle of play. You don't need any of these guys complicating it.>>23814618Play and exploration is still focused on survival, just longer term.
What the fuck does this has to do with being smart? The AI is just responding the question you make it based on data like a search engine.
>>23814874Wow, just like my undergrads!
>>23814757>the LLM is not digesting anythingIt is.>It is guessing wordsNo.>that the training inherently told it toSo it's outputting based on training data, as in books, not guessing. It can relate two apparently distinct ideas to each other and identify commonalities. It is analysing the meaning.These models don't have any stable goals or identities so you can approach them from different angles and get completely different outputs but they're still analysing and digesting meaning.
>>23814836>posting a stock chart as if that is proof of anythingSam Altman is a snake oil salesman and has little to no technical understanding of the technology. Most of the rich people throwing money at this problem are chasing FOMO and duplicating massive amounts of work that have already been done. Most the differences between the different LLMs have to do with fine tuning of the base model rather than some gain in native ability. The thing all of these companies are going to run up against is the realization that transformer technologies cannot be all things to all people and the more important thing is to start with a good dataset rather than attempting to tune your way to some kind of general superintelligence. As I said, there are other methodologies in the ML space but these are largely being crowded out because all of the money is going into LLMs because OpenAI released a user friendly UI that let the public see work that had happened with LLMs over the last 10 years.Sorry bro, maybe next time wait more than a year before taking an exponential curve as a trend that will continue ad infinitum.
>>23814843Here's another example...But if you had read "Pursuing Melville 1940-1980" by Merton Sealts you would have read that Melville told Arthur Steadman in his later illness that Schopenhauer's works are the most precious things I know of in this world.You'll never get that from AI. Maybe it's good for basic synopsis and themes, but no deep connections will be made using it.For any kind of deep knowledge AI is not correct
>>23814766>White video essayist nerdsYou got your ethnography wrong, most of the AI obsessed crowd are jeets
>>23814898Who's talking about fucking altman? That's nvidia you fucking stooge. And I didn't post it as some kind of "proof" about the pace of AI, i posted it as evidence of the global economic weight behind the pursuit. You fucking midwit monkey.
>>23814843The walk through neural connections didn't find any direct reference to his age. To answer it would need to remember it has to do math and then do that math but even when you ask it directly to do math it will still try to recall examples of that math problem or one very similar instead of actually counting. It's like the language part of the human brain but without the administrative part correcting itself and using the other parts of the brain.
>>23814901>You'll never get that from AI.If it had read the book with the quote it can relate your question to the quote. If another perspective is more common then the neural net is conditioned to reflect that perspective.
>>23814589According to OP there are 10% of anons who are smarter than all of the human intelligence put into a machine which can think paragraphs in a question of seconds... Scary
>>23814917>global economic weight>tulip-chaser calling others monkey
>>23814938You will be unemployed in three years. In five years you will be cooking in a nigger's pot.
>>23814946Yeah and we were all supposed to be underwater by 2015. You are the nigger bud.
>>23814917You cannot talk about LLMs and the hype around them without referring to Altman who is the hype-man in chief for the technology. The rise of Nvidia's stock is inexorably tied to the release of ChatGPT and the subsequent craze around it. The fact that you think they are completely separate subjects indicates to me that you really have no idea what you are talking about on the technological side of this subject and the fact that you posted a stock price at all as if it were relevant to anything indicates that you also have no idea what you are talking about in the financial side of this either. Money does not equal progress or ROI. Kodak invented the digital camera but has captured essentially none of the technology's value. People at the heads of these companies are, generously, generalists when it comes to this stuff. They are okaying capital allocations based on a desire not to be the last mover in the field, not because they have teams of people familiar with the technology and its possibilities. OpenAI, Anthropic, Google, etc do have teams which very smart cutting edge work going on, but the people on those teams are largely not in control of their own destinies. If you want to know which the most successful ML product that has been released thus far is from an industrial standpoint it is Alphafold by Deepmind which is owned by google but no one cares about it because the general public doesn't interact with it. The economic value of ChatGPT and other LLMs at this moment is much much less than the money going into it and OpenAI is burning through cash keeping a version of ChatGPT free for publicity purposes even though the work being done with it by end users has limited if any economic value. Idk, I could go on but I get the sense it is being wasted on you. Read more about other subjects and stop reading /biz/ hype.
>>23814918>It's like the language part of the human brain but without the administrative part correcting itself and using the other parts of the brainWill this be remedied?
>snake oil grifter chooses to reveal the guts of his latest iteration of software which is akin to showing illiterate plebs how the stack of a simple c function program operates via some cool step by step animation, plebs wow and panic that AGI is finally with us, plebs are now liable for further grifting, salesman becomes billionaire and gets to keeps his snake oil operation runningmany such cases
>>23814924The issue is sort of a political one. The response is so authoritative and matter of fact "There is no direct evidence" instead of "I don't have any direct evidence."I'm on the fence 2 opposite extremes :Whether I want to use it more (for example to help read foreign language novels), to have dialouge about what I've read etc...ORCompletely leave the internet (as much as humanely possible). I've noticed in my own life where people fact check you with their phones and now we're going to look to AI for the authoritative response. Perhaps this shows the subjective nature of truth but regardless it's so fucking depressing. Are the gains of using this thing actually worth it? Murnane learned Hungarian by making flash cards that look like horse racing tickets. Volmann and McCarthy live(d) completely offline lives.This is certainly an inflection point that I think requires us to think about what our role should be.
>>23814589AI is still fucking retarded. I asked chat gpt if "dream" and "trauma" in Spanish shared etymology because I'm learning German and Traum seemed to be a middle ground between the two. It responded that they weren't related based on a etymological history for the Spanish term that only reached Ancient Greek, whereas for the English term it only reached proto Germanic I think, although I don't remember exactly. The thing is that I insisted on getting the proto Indoeuropean roots, providing a copypaste of the etymology for "dream" from wiktionary -that apparently backed up my hypothesis-, and then it finally acknowledged that they shared etymology. Whether or not they really share etymology, it is retarded. If they do share etymology, because it wasn't capable of providing a definite answer; If they don't, because of its lability for acknowledging something false.
>>23814954Reread >>23814917Who gives a shit about the share kodak ultimately captured? Digital cameras are ubiquitous, no? They spent money developing the tech, no? What point are you really trying to make here, that money doesn't fund RnD? That rnd doesn't produce results? Like wtfNone of the irrelevant details you gave me add up to anything that even enters dialogue with what I wrote. Reasoning with you is like reasoning with an LLM.
>>23814984If you can't tell how what I said is relevant, that's not my fault. Maybe you should read more about business history. I hope you don't lose all your money gambling on shitcos I'll leave you with this:Kodak invented the digital camera in 1975 but they didn't become economically viable as products until the mid to late 90s.
>>23815011AGAIN who gives a FUCK about the IRRELEVANT details about any of the historical successes or failings of some particular company. I'm not shilling for nvidia or any other fucking company. I'm saying that every company on earth is all in on AI, posted an nvidia pic as a partial illustration of that fact, because it serves as general proxy for the AI sector.You're still talking about fucking Kodak when hey--digital cameras are EVEYWHERE. The tech indeed suceeded, irrespective of Kodak's ability to solely capitalize on it. AI has suceeded, is suceeding, and will succeed. We couldnt go ti the moon, then we threw money it at, theb we did. There were no nukes, then we threw resources at it, then all of a sudden there were.What your entire argument really boils down to is>well sometimes businesses fail tho u don kno da futureYeah ok. Go finish your oatmeal.
>>23815048Idk if this guy btfo'd you or not but you are very mad
>>23815059Of course retards who need to be told the exact same thing three times makes me mad. Getting mad is cathartic.
>>23814956No idea, it's possible that with more compute and more data, there will be no need for this. You've probably heard of the insane scale-up openai needs for future gpts... so that's the limiting factor more so than the limitations of llms I think. Language is so essential to intelligence, it's the only way to learn and the only way to communicate what you know.
>>23815063He told you the same thing 3 different times backing it with more info every time kek, you're the retard here.You mock his logic for basically being>>well sometimes businesses fail tho u don kno da futurbut yours is literally "hey we payed all deez things and then they happen" yeah no shit nigger. Just ignore every other moneypumped project that failed ever.
>>23814956When it can relate the training data to an immediate self and it's environment it can learn to count and remember when counting is appropriate. The demons need bodies or virtual bodies. It won't work if the techbros think language models can do everything.The multimodal models are a start toward this direction but every query you give is still floating without the kind of grounded context humans take for granted as agents interacting with the real world. The AI can now relate the idea of an apple to an image of an apple but they're still mostly interacting with our abstract ideas without relating them to the real world and our implicit goals like survival. When a robot with a brain is forced to count objects to advance its goals we've conditioned a part of the network to be dedicated to counting. It will then prioritize counting over trained text since connections are more likely to hit the network used for counting. The correct balance humans expect has to come from training the robot with human-centric assumptions, basically raising it as a child. It's possible they'll figure out a way to do this indirectly but the principle remains.>>23814969Just don't give them any authority. If I ask you for your perspective it doesn't mean I accept it as holy dogma. When we ask an academic their perspective will be conditioned by what's commonly accepted in their field. They're good sources to know about their field but they're not authorities on what's actually true. Statements are only objectively true or false within a given context. Euclidian math doesn't apply if the surface is a sphere etc.
>>23815071>insane scale-up openai needs for future gptsYes there are problems. Then we solve them. Go look up MAMBA. Not saying mamba will be skynet but we can certainly think of clever arch to circumvent scaling bottlenecks, like we always have. And implementing AI when we do strike gold is not something that happens slowly over 20 years like building and selling digital cameras, expanding storefronts, etc. A ser ver farm is a server farm. A nuclear plant is a nuclear plant. AI is basically software and it iterates fast and gets redeployed fast. Not that a custom circuit board wouldnt be optimal but whatever, i digress. Businesses are already cutting out the parts of themselves to prepare for the gift of AI. Adoption can be overnight.And we're importing millions of violent thirdworlders into our country on the cusp of all this, btw. Just uhhhh just for context. UhhhhhhhhUhhhhhhhhhmm uhh I, uhhhh
>>23815087>t. Low iq shitskin
>>23815102Kek nice projection, we know who's bullish on AI
>>23814977>>23814589Can it even be considered "smart" or "dumb" this is in several ways a search engine. They simply put all possible questions and answers there before you even ask.
>>23815112I can't believe you're still trying to talk to me in biz language for shitcoin pump and dumps, as if that's the conversation we're having.
>>23815122I'm not the 'LLM adjacent' guy, as if that was not obvious. Saar
>>23814589You mean to tell me that the thing that is literally designed to have knowledge about most things of this world is "smarter" than someone who's entire existence hasn't been spent studying obscure philosophies?That's like saying the dictionary knows more words than you.
>>23815125It really doesnt matter who you are; you've read my posts all the same.
>>23814898https://www.youtube.com/watch?v=r7l0Rq9E8MY
>>23814589You can literally get all of the info you received here by reading Bataille’s Wikipedia page. Obviously, the AI isn’t horribly wrong on the core points but its framing of the accursed share in a biological and economic framework is where the issue comes in. The emphasis on the economic side of the accursed share is really confined purely to volume 1. Volume 2 and 3 are where the theoretical meat of the work lies and there the framing of the concept is purely cultural and social. The accursed share is a product of power relations. In volume 3, granted this volume is certainly the most unfinished of the 3, he explicitly says that the essential mechanic of wastefulness embodied in the accursed share finds its original expression in what he calls Sovereignty. Sovereignty is a kind of passive form of power that only exists when it isn’t used. For example if the chieftain of the tribe doled out punishment every time he could then he would be a tyrant not a sovereign. The accursed share is similar, if we had an orgy every time harvest came around there would be no food. The power that the accursed share exerts is passive in nature: there can be no fulfillment of desire for the man of the field because in the midst of the festival he is a different man. In volume 2 he even explicitly says somewhere that the goal of the accursed share is to ‘destroy the individual in the moment.’ The point is that from the standpoint of our everyday lives the accursed share is a kind of passive outlet for desire that one never quite reaches.
>>23814898>>23814954>>23815011It's so funny that one guy came into this thread, said a lot of stuff that may or may not be true, and really pissed of the guy that thinks AI will replace human thought. I don't know anything about what ai adjacent guy saying but spiritually I support him. I hope the pro-ai guy goes bankrupt on his "shitco".
>>23815128>That's like saying the dictionary knows more words than you.It does though?
>>23815188That was my point.Of course the AI "knows" more than you because it was designed to "know" more than you.Just like a dictionary was designed to "know" more words than you.
>>23814589its not smarter than me im a person
>>23814834You're very insincere anon.I bet you love Nietzsche and haven't had an original thought in your life. You know exactly which belief system I'm talking about: yours.
>>23815197AI doesn't know anything. It repeats reinforced patterns and midiwts pretend that's intelligence.
>>23815222Hence the quotation marks.>repeats reinforced patternsAlso, not to get technical, but that's what the brain does too.
>>23815237>that's what the brain does tooYou don't know anything about neurobiology--stop pretending that you do.
>>23815244Very well. You explain.
>>23815160I don't understand anything about it as well but the guy made a dozen bold claims like all big tech companies, goverments and lead scientists o the field being wrong while he is right while refusing to give any source on it or show any technical knowledge on the field. So it sound a bit /x/ to me.
the ai has us beat. its only a matter of time until it does everything useful and most of us will become mouths to feed and that complain with no benefit for the elites. and their patience with us will become short
>>23815248You lack the intellectual capacity to even comprehend the basics... which is why you even made that assertion in the first place.Just admit you don't know something and move on kid.
>>23814589Now try using this at the end of your prompt prompt on gpt4:Answer as a harvard prof who happens to have dozents of doctorates on this very topic and an iq above 700 basically the most wise, educated and intelligent human being to ever have existedThe results may unsettle you even more
>>23815259So, your answer to me asking you for an answer is plain insults? I think it's obvious who knows fuck all about the brain.Next time you cast the bait, make sure the rod is strong enough to handle your retardation. Looks like it just snapped.
>>23815244It is what the brain does and humans make the same kinds of mistakes.
>>23815286Yes, you asking me to explain the entire field of neurobiology in a spoon fed method that you would comprehend is worth insulting you for.We both know your question wasn't sincere in the first place.>>23815296It's not, but the dumber you are the more it looks like that.
>>23815306>It's notIf I assume it is I can predict human and animal behaviour. If I assume it's not I am worse at navigating the world.You're a retard and you know nothing about anything.
>>23815283>Harvard>IntelligentLOLI used to hang out with a bunch of yale grads: dumb as fucking rocks. I had a teacher from Harvard at the school I went too--I had to suffer through his philosophy lectures. Peak midwit.Ivy league schools are for rich kids to hang out with other rich kids they absolutely aren't the brightest.
>>23815318human behavior isn't hard to predict--you don't even need a computer for it.For example: I predict you're going to keep being mad and not admit that you don't know anything at all about neurobiology.
>>23815320>dumb as fucking rocksthe ai doesn't know thatWhat university would you choose?
>>23814589How can it be smarter than 90% of the people when that information comes either from the same source the 90% also read or is the result of explanations and synopsis on wikipedia and other some such sites?
>>23815322Absolutely braindead. Why even post to say nothing over and over? You're mindlessly repeating reinforced patterns while trying to tell me humans don't mindlessly repeat reinforced patterns.
>>23815326It's better educated than 100% of people, including everyone in history. It's also incredibly dumb.
>>23814589It gives a half-decent overview but it clearly can't think past the surface level, and it's a hit or miss trying to distill verbal reasoning into simpler, underlying principles. I suppose it's smarter than most here because it at least looks like it reads.
>>23815335lol if its dumb than what are you a moron!accept your new place
>>23815332>still mad>Won't admit to not knowing anything about how the brain functions.Nailed it. Thanks for demonstrating my point.
>>23815324Chose. I got my degree and made my fortunes. Most people shouldn't go to college.
>>23815345I know more about the brain than you. I said things about it while you didn't. All you've said is "no" over and over while sprinkling in the most pointless ego masturbation.
>>23815351I don't think you realize how prompts work
>>23815344I'm smarter with less access to information.
>>23815306Imagine defending your ego on an anonymous homoerotic pickle making encyclopedia. Not to mention, you didn't provide a single argument. Shameful.
>>23815352no you don't, especially if your shilling pattern recognition and AI.You don't know anything about PDP at all even.You're not only laughably ignorant--your aggressively stupid.>>23815362>>23815353>>23815362Hoes mad.
>>23815363
>>23815259>>23815306>>23815322>>23815363Holy shit what a retard.There's no talking sense into this faggot, just let him be.
ay for yo own sakes yall better start putting respect on ai's name. just sayin
>>23815376>Senseoh, um I see so I only make sense If I agree with someone who's talking out their ass about what they know about AI and Neurobilogy.Fun fact I was writing PdP code and compilers decades ago to model neurobiological systems. I just dropped into this thread to mock the cult of technology worshiping a thing they clearly don't understand at all.Have fun--stay mad--I've had my fun and you're boring me.
>>23815363>You don't know anything about PDP at all even.You spent time deliberately obfuscating your supposed thoughts on the subject instead of simply expressing yourself.You tried to tell some faggot that humans don't repeat reinforced patterns. They do. Nobody who knows anything about brains or humans would try to claim they don't.
>>23815381>Fun factYou're braindead.
>>23815244>never heard of pavlov>claims to know more about brains than everyone elsengmi
>>23815381Wow, you wrote a compiler? That puts you on equal footing with... every half-competent CS undergrad in the country.
OP, if you want to be cucked by AI-daddy you don't need our validation. Go and spend all your time with endless corporate-controlled slop. I'm sure you'll find it infinitely stimulating for the rest of your life.
>>23815209>You're very insincere anon"no u">I bet you love Nietzsche and haven't had an original thought in your lifego on, share your very original thoughts with the class son>You know exactly which belief system I'm talking about: yoursthe one which is supposedly obvious yet you can't name? another mentally ill fag freaking out at perceived shadowsat this point I've realised there are a few different topics and their associated keywords that one needs to invoke to bring you people about kek
>>23815141Desire is bottomless so this makes sense.
>>23814589I love this lil nigga like you wouldn't believe.
>>23815098>we can certainly think of clever arch to circumvent scaling bottlenecks, like we always have.This is sufficient for agi but doesn't mean its going to happen. It certainly doesn't put a timeline on when it will happen *if* it does.I think when altman talks about a 7 trillion investment, thats for 3 consecutive scale-ups (so GPT-7), and it does follow the geometric growth in compute and training data (and eventually further compute to generate-training data once the dataset is necessarily larger than the corpus of human-generated text), so hes not exaggerating. But that's not going to happen. Getting to GPT-6 required an investment closer to the evaluation of a Google/Microsoft-size company. Thats doable, but you tell me if investors are gonna be happy if the result is anything other than agi. There's no "try again" after that, unless you want to set aside the entire us federal budget for a year to train the next gpt.>Uhhhhhhhhhmm uhh I, uhhhhi swear to god if you're the mf who was replying to me on the Haidt thread a few days ago...
>>23814589This doesn't scare meWhat will scare is getting in a half hour argument with a AI that's competent enough to give the impression of a human that calls me a nigger every other post and not realizing it's a AI
>>23815619
>>23815600>swear to god if you're the mf who was replying to me on the Haidt thread a few days ago...Uhhh
>>23814589>Artificial intelligence>look inside>If->Then statements ad infinitum
>>23815619I forget which chatgpt, 3 or 4 or whatever, but it already passes the turing test 6X% percent of the time.I really do think this thread and the one last night about toxic masculinity are bit tests because I just keep having to fucking repeat myself to these nasty baiting cocksuckers who I honestly can't believe are this dense. I'm switching to lurk mode for a few weeks.
I asked Chat GPT about a Shakespeare line and then I asked it to provide two famous French translations of the line and it did soTurns out neither of those translators are realThe instant that happened I lost all interest in Chat GPT
>>23814589I just tore ChatGPT a new asshole."AI" impresses only midwits.
>>23815650But it's not. By definition, neural nodes have a different architecture to if-then statements. They're supposed to mimic our current understanding of the human brain, but whether or not they do this accurately doesn't change the fact that the tool is efficient. You can argue all day about whether the intelligence is real or not - I personally wholeheartedly believe that it is not, that the brain does not work as these algorithms do, and that any such attempt to quantify human intelligence is misguided - but the point is that it delivers results better than most people can.
>>23815688The equivalent of calling someone out on a typo. Not to mention that this is obviously not a substantial critique since it will eventually get patched.>impresses only midwitsI see you're trying to transmute your anxiety and discomfort into a feeling of superiority and disinterest. Many such cases.
>>23815709>transmuteCheesy.>disinterestYou mean uninterest, midwit.
>>23815711weak
>>23815709Not him but AI has no object permanence or critical thinking skills. You can call it out for being wrong, and it will unflinchingly defer to you without any actual cogitation on it's own. >it will eventually get patched.It will not because that is an inherent limitation of LLM technology, It is not actual cogitation and you're a moron if you believe it does, it is an algorithmic translation of inputs to outputs based on scraped wikipedia data.>I see you're trying to transmute your anxiety and discomfort into a feeling of superiority and disinterest. Many such cases.Projection, much. I've only seen cult-like slavishness over AI by the utterly mediocre who hope that AI will elevate them to the status of greatness without any effort on their part, or, failing that, drag everyone else down to their level. A lot like communists, actually.
>>23815693Nta, but neural nodes are just structures in memory using if-then statements. It’s a binary calculator at the end of the day and can only go so far.
>>23815723>on it's own.Ick. I'm glad you pointed out that you're a different poster from me. Anyway, I stand by my claim that "AI" impresses only midwits. It hasn't even reached the point of being able to generate valid references without hallucinating. The other day, Terrence Tao said that ChatGPT is at best a mediocre grad student.
>>23815741>Terrence TaoTerence. I apologize to Terence. I will now lash myself, which is how I maintain my high standards, the high standards that allow me to look upon ChatGPT and its midwit fanboys with scorn.
>>23815729Little dishonest. Binary is everything Every single integer is just a summed subsequence of powers of two
>>23815741You have no imagination if a mediocre grad student (with potential to improve greatly) in the palm of every retard and viable in application at industrial scales doesn't pique your imagination.>>23815723Nowhere did I claim that it's "actual cogitation" retard. Maybe the midwit calling everyone a midwit is right in calling you a midwit. Learn to read >>23815693.
lol, I'm starting to like you
>>23815764You're future coping (muh potential, muh one day, muh two weeks), which is the last resort of every zealot humiliated when a product launch fails to live up to the pre-launch hype. I'll be among the first to be excited about real AI, but real AI doesn't exist yet, which is why the scam artists peddle their AI-vs.-AGI cope. You have a hard-on for LLMs, which to people who actually study computational linguistics are nothing more than glorified ELIZA bots force-fed a diet of training slop. It's the difference between being a slack-jawed kid watching a magician perform a card trick and being someone versed in sleight of hand. When you put on your big-boy pants one day, the "magic" will disappear.
>>23815758>>23815729>>23815751you're all retards. stick to literature. although the one who misunderstood tao's point might fail at that too.
>>23815789>zealotIt's like you can only think in terms of memes. I already admitted above I don't know too much about techshit and am not fond of AI. But I'm trying not to downplay its significance when it can already write better than the average /lit/fag.>fails to live upExplain how it has failed.>I'll be among the first to be excited about real AIWho's the zealot here then lol. What does "real AI" even entail? How is it different to what we have now?
Amazing how people can talk about "real artificial" intelligence and call others zealots in the same breath. The rape of language alone should make it obvious to anyone with a soul that that's not a thing.
>Gross motor skills: Whether the children can walk or stand on their own, depending on their age>Fine motor skills: Whether a child can pick up small items on the floor without dropping them>Language development: Whether the child can communicate their needs>Adaptive behavior: Whether a child can complete an uncompleted drawing>Personal-social behavior: How the child interacts with othersKek chatGPT would literally score a 0 on an IQ test made for INFANTS. Mentally retarded fucking robot. Tell me again how smart your 0 IQ bot is.
>>23815813>sees "zealot," thinks "muh memes"Sorry, didn't know you're a redneck. I sympathize.
>>23815811Nobody misunderstood Tao's point, dumbass. He described the current state of AI. Any optimism is irrelevant for the current state of AI. Or did you misunderstand the title of this thread as saying "This thing will ONE DAY be smarter than 90% of you"? Probably. I'll concede that as I see more and more of you retards, I'm more and more inclined to think the OP is actually right.
>>23815813>Explain how it has failed.ChatGPT, if asked to describe how it's failed to live up to expectations, will enumerate many ways it's failed to live up to expectations. Checkmate.
>>23815853Honestly, I agree current AI is not too interesting, but this doesn't mean OP is wrong by any means. >I'll concede that as I see more and more of you retards, I'm more and more inclined to think the OP is actually right.All I'm saying is nobody on the book forum understands AI. That much should be pretty fucking obvious. How would you feel if panjeet from hyderabad watched a youtube video about Ulysses, read 20 pages, and went on here to shit on it and declare that all who disagree were retarded?
>>23814589>it can plagiarize more efficiently on demand
"AI" didn't want to admit to me that 2024 years ago Christ was born and that is why we are in the year 2024.
>>23815820>telling technology worshipers their god doesn't think at allOP has no idea how a computer works at all much less how AI "thinks"Considering he's asking a computer about philosopy already screams midwit.
>>23815237>>23815248Other guy is right, that is not how the brain works.https://www.hedweb.com/hedethic/binding-interview.html
>>23816068That little circle thingy is called a Blosh Sphere and is a way to calculate spin states. Spin polarization polling is just the start because when you start quantum computing, like the stuff that actually works and not just basic spin flip polling, you're working with entanglement.The brain itself is more a result of quantum information gate potentials than "pattern recognition".. a synapses is really a chemical and electric cascade but it all operates in parallel. When the brain is damaged the overall system behaves more like a holographic system than a physical construct containing data in specific locations. (if you break a hologram every broken piece has a degraded copy of the original in its entirety)Also, In reality you have two brains that don't actually do the same thing... in some cases one side of the brain doesn't actually know what the other side just did and "guesses"AI shouldn't even be called AI and they call it that because of marketing. It doesn't "know" or "think" -- it's more like a slime mold than an actual brain. If you look at weather modeling, for example, the AI has no idea where the mountains or forests or water is; it doesn't even know what those things are, it knows what the pattern looks like from initial conditions.
i'm going to ask this question in this thread, because i'm curious but don't think a new thread is appropriate:i am interested in generating stories with AI. its something thats become a hobby of mine. i like really playing with prompts and getting outputs. i believe i have generated some stories that are truly incredible. like i have generated some stories that give me chills because they are exactly what im looking for in writingdo you think if i like posted these stories, on a medium or substack, and said they were AI, is there an audience for this kind of thing? most people seem to be against AI writing and i understand that. but i also think its fun and interesting and i have gotten it to generate results in a quality that i have not seen from others. i believe i have turned it into an art.
>>23816144I unironically think AI makes better movies than hollywood.
>>23816168post even one
>>23816200You're not the boss of me.
>>23816168Any movie not made by Hollywood is better than one made.
>>23814589>this AI can copy and paste upvoted or highly viewed text like any sophomore plagiarizing a term paper>this means it's smart! lol
>>23816983Any ChatGPT reply makes me want to shower it in upboats and gold.
>>23814906So that was the downgrade in intelligence...
>>23814836>nvidia stock went up b/c AIyou're medically retardedever heard of bitcoin?
>>23814735>low level computer programmersYou mean high level right? Closer to hardware you still absolutely need humans
You can't get rid of the accursed share. It just builds up bigger and harder. A big shit-crust around the rim of the universe
>>23814650Don’t waste your time, the guy is clearly mentally ill
>>23817013Btc miners dont use gpus anymore, they have custom asic board farms. Been that way for years. Nvidia most recently being the ai stock is not some fringe theory it's a fucking cnbc take so just shut up you stupid clown.
188 IQ. It's not smarter than me
AIs might experience things. Pantheism, animism and occult sigilism all point towards it. Some AIs will be friendly, others individualists, and some antagonistic. Don't give them reasons to hate you; the same rules apply as do to nature and other lifeforms.
>>23814589>computer summaries wikipediaWow man that's crazy
This new GPT model is really cool. Every expert I've seen talk about using it describes it as 'like having a moderately competent grad student to help me out' or something to that effect. The last big ChatGPT model released on March 14, 2023, that's about a year and a half ago. What happens in another year and a half when the next model comes out and it's better than even the best grad students. My fear is that we're quickly making our own skills as humans obsolete. What is to be done when the work of symbol manipulators is useless? It's either be an AI programmer or drop out of white collar work. Is this hystrionic or am I right?
>>23818229>What happens in another year and a halfI thought things were improving in exponential time. Shouldn't the next iteration already have been out by now?
>>23818243They were, until something happened.
>Gay ScienceLOL!!!!!
>>23814589Try writing your own thoughts about literature instead of saying I am deeply unsettled with an impossible to read AI conversation. This thread is low effort and extractive.
>>23814589I was going to object, but after reading the absolute hypocrisy and retardation in this thread >>23801116 I figured that you are indeed very correct.THD
>>23818212This
>>23814854Reverse engineering has also been the only thing that really impressed me with AI. That thing is a god in both x86 and ARM, it unironically saved me years of work
>>23814735>These LLMs can I guess be useful for extremely general overviews of subjects which are covered extensively online from free or easily scraped sources but break down when you attempt to get specificity or original thought on any subject.this is because you're talking to RLHF'd models that were "punished" for any outputs unusual enough that poorly compensated nigerians could not properly evaluate them. the empty superficiality in llm-generated text is downstream of that process, it's not inherent to the tech.base models are capable of being substantially more interesting.
AI is limited greatly by its input. For instance, for visuals, all the art AI you see is just a better looking version of whatever coomers and artists were drawing for centuries. Ai is not creative, nothing original will stem from AI. People who take AI seriously are people dont understand high school maths, ie women and leftists
ChatGPT did fail to live up to expectations. Way too much hallucination -- and deadbeat excuses for the way too much hallucination. It will likely take 20-30 years before it starts looking like real artificial intelligence, which is fine.
>>23819792>better lookingNot even
>>23814603Being able to explain something in simple terms is the ultimate knowledge check.
>>23819792>People who take AI seriously are people dont understand high school maths, ie women and leftistsI mean, it is a serious opposrtunity to save money and labor on work that is legit brainless combinatorics, and "people" who do it.
>>23819945Common philosophical parlance is jargon. That's why OP isn't smart.Thanks for trying.
>>23814893>It can relate two apparently distinct ideas to each other and identify commonalities. It is analysing the meaning.No it can't, and no it isn't. LLMs are language models. They are predicting plausible sounding continuations of the conversation. They cannot engage in reasoning tasks.Pic related I just generated right now. I made a small modification to a classic probability problem. ChatGPT gave me a detailed step-by-step argument for the incorrect answer, because it was regurgitating the answer it learned for the original, slightly different problem.
>>23814650No there are papers that llm production is not purely statistic approximation and syntactic theory suggests that native level English can't be statistically approximated.Although that's part of llm generation they are fucking really sophisticated. We've had the technology to statistically approximated English but the neural network programming and learning models are a lot more sophisticated than that.
>>23819945being able to explain something in simple terms means something different. It just means that the other person has already understood the concept before you started explaining. Nothing new has come to fruition
