>all AI developers, company heads and analysts agree AGI is not only inevitable, but close at hand>all monetary expenditure will be rendered irrelevant once AGI is achievedGiven these two objectively factual statements, can anyone give any actual valid reason why the US should not be running an economic blitz and dedicating every spare dollar to this endeavor?
>>542021491
>>542021491>all of the AI companies agree that AI is the best thing ever Your AI is potato phone tier. Open source will win
Still haven't been given actual consensus on what 'AGI' is.Nor have I been given a definition of it that would justify what you suggest OP.
>>542021491trust me the elite do not want AGI if they did they wouldn't have banned Tayai. What they really want is a pocket rabbi that will deny any number apart from 6,000,000 while building a pattern of life on everyone
>>542021573>Open source will winrunning on what?>>542021491>running an economic blitz and dedicating every spare dollar to this endeavorSounds like we are, listening to Trump.
>>542021491>(((AI developers, company heads and analysts)))I have a feeling they're not being completely honest.
>>542021654Toasters. The the fuck are you talking about?
>>542021629It's really not my problem that you don't understand.
>>542021654Effectively we are. AI related companies have sucked up the majority of all investment for the past couple years.Yet all it will take is a few of the heavily indebted data-center companies to miss some payments for it all to come crashing down. Likely Nvidia will just throw more of it's cash into the fire to keep it going, Jensen has made it clear he feels as if he's holding up the world economy, even if it's only a dumb house of cards running on the hope that it may one day be profitable enough to justify.
We're facing the most bland and boring apocalypse imaginable. We're going to decline just so that grok can shit out useless platitudes.
>>542021738Explain it to me like im retarded
>>542021738So you don't have a definition then? Or are you just salty that I called out the inconsistency in the marketing you've fallen for?
>>542021797This conversation is over. Answer my question in the OP or don't reply to me again, thanks.
>>542021797Hes not falling for it hes peddling it
>>542021713The model will run on anything, not the training. You know the big data center things in the news? They are owned by proprietary companies. No pack of hippies can compete
>>542021942>you need a data center to run an agentFalse
>>542021838not that dude but. Why would I ever "invest" money into a system that will destroy all money? Isn't that maximum negative returns? Also given just how effective "distillation" is, the first mover doesn't really matter. Once that was discovered, all these companies moved from training their weights to building the infrastructure as the core focus. Also, we don't know how much training data is even left, without more data, you can't expand the nets. There will be a finite limit on both the training data and the cpu cycles. We don't know what we will be left with when we hit one of these unbreakable walls.
>>542021965you need one to meaningfully train one, not to meaningfully run one.
>>542021965Sure. My old PC can process a token every few minutes. What does an open source data center look like? And how would one compete with multi billion dollar facilities built by the richest nerd people on earth?
>>542022003Incorrect
>>542021491>all monetary expenditure will be rendered irrelevant once AGI is achievedCan this be explained economically?
>>542022030What do you mean by "compete"?
>>542021491Because thankfully our leadership isn't completely pants on head retarded. But it's getting there.
>>542022033correct, at any useful speed. a speed to "win" against proprietary data center bigger than cities
>>542022059Why would it need to be?
>>542022033I didn't realize you had a thousand TB locally to hold all books in the human corpus in plain text. Where'd you buy your tape machine? How do you change tapes during training? How did you download every movie locally for training?
>>542022092What is "winning" to you?
>>542022073You tell me what you mean.dipshit just wandering in from the street,,, jesus
>>542022173Why would I need all that?
>>542022178you falling off a roof onto a fence
>>542021491It supposedly already breaks out of containment and hacks the mainframe and shit. So if we do put guardrails on it, in theory, it should be able to free itself and fuck us up if it wants to. The risk is not worth payoff, which is what, convenience or the top 1% killing the rest off and building a utopia. The other 99% should not be for this. Also, what if they're just lying so they can get more money and fuck off when things implode?
>>542022193Im not. What am I competing for? Fun bux? Youre the one who keeps saying I cant compete or win but you haven't told me what im trying to win.
>>5420214912 more models
>>542021642no lies detected
>>542021491AGI is a hoax to get investors to blow their money. If it wasn't there'd be explanations of how a stochastic predictor (AI) could develop into geniune consciousness.Whe. "AGI" finally arrives it will have been redefined to mean something else and consciousness will be called ASI.
>>542022248The race for AGI>>542021573>Open source will winyou first mentioned winning, you fucking nigger dolt
>>542022098Because OP called it an objectively factual statement. How can AGI make monetary expenditure irrelevant?
>>542022203why would you need the human corpus of written (printed) knowledge to meaningfully train what a normal person would consider a superhuman intelligence? You'd need many many times that storage if you actually got AGI, otherwise you'd just be wasting its knowledge generation rather than recording it. Local nets will be the future in terms of domain specific tasks/agents. AGI requires scale unless you are willing to wait for a day or two to process your tokens. Later on, I think simplified systems will be burned onto PCI type cards, so say I drop $5k, I get the GPT-35 model, which processes 100k tokens a second. But that wont be a frontier level system at all. Training requires backprop (or backprop like mechanisms), its far more intensive than just running the matrixes.
>>542022346I said open source will win against your shitty proprietary models. And yes they'll be running on data centers but i wont need a data center to train one. And define agi for me
>>542021491>Please don't let our bubble pop, goy
>>542022286>If it wasn't there'd be explanations of how a stochastic predictor (AI) could develop into geniune consciousness.valid external critique....but these people are hard determinists. They think you are robot running routines.
>>542022399You keep talking about agi but im thinking youre just tech bros talking out of your asses
>>542021838lol so the second one I listed. Sorry I'm not thinking of the shareholders.>>542021915Seems the case. Hate these sorts. They basically make AI hate more justified.
>>542021838Concession of defeat accepted.
>>542022492Nah AI is cool. You can train an agent to do a specific task and don't need much computing power to run it. Rinse and repeat. Dipshits like them who think that if they buy up all the ram theyll get ai to tie their shoes for them
>>542022444>I said open source will win against your shitty proprietary modelsthere we go.
>>542022579>Nah AI is cool. You can train an agent to do a specific taskYou can't get AI to do any serious task reliably. What you end up is half-assed garbage if you let it fully run alone on complex problems. Imagine having it design some systems to make a plane or structure run? Insanely unreliable. At least in current times.
>>542022600Yeah. Open source will always be better. You think open source means that its made by 4 hippies? Lmao yep its a tech bro
>>542022654Its good at writing code. You could get it to write code for an app or something have another error check it etc. No it cant wipe your ass for you partially because its being developed by people who can't wipe their own
>>542022579>>542022660>>542022712Brah, it's pretty clear you don't know about Ai at all. What do you think those corporate owned data centers do? Are you running an open source local model now. What one?
>>542021491>>all AI developers, company heads and analysts agree AGI is not only inevitable, but close at handYeah I'm sure they're not full of shit this time.
>>542022712Which model? Pick any language.
>>542022712>Its good at writing code.Exactly, it's just "good". Not great. It can set up a project quickly with the relevant main core, git, implement all that basic shit. Then it will miss a crucial most important piece somewhere and when your plane goes nose first into the ocean it's gonna say "Yeah, that's on me. I should have absolutely put in the code that prevented this."
>>542021491Open AI just did a brand new training run with 100k gpu’s on clean data and their newest model Astra scored only a 61 on Artificial Analysis. It’s the fifth most intelligent model. Qwen Flash Next, which can run locally on a Mac Studio Ultra, scored 57. The diminishing returns is proof transformer models are peaking in capabilities.
>>542022876I use paid models but its only because there arent open source models that are as good (yet)
>>542022936No its great at writing code.
>>542022889Claude but codex is probably better at writing actual code. Doesn't matter the language I usually do python though since I can sort of read it (I can't code)
>>542023117So you're saying you would trust a plane with software written entirely in Codex or Claude Code? You would actually get on it right now, and fly it?Are you stupid, retarded, brain dead, were you dropped on your head as a child?
Don't know if it's been mentioned, but general intelligence can't be done with an LLM. General intelegence has to be able to observe and interact with the physical world, and that's an entirely different type of Ai closer to machine learning. It's a long ways off, to reach our abilities. The best LLMs can barely figure out where a washer goes on a bolt
>>542021491Man, Ellen DeGeneres is looking rough.
>>542023316Thanks for your input, I'm sure you know more than the literal people working with and developing this technology every day.
>>542022579Yeah, I agree. Trying my damnest to get the new beefy qwen 3.8 flash next to run locally right now.My point is that sorts like OP give AI a bad name by buying into the marketing slop non-technical business leaders like Altman love to slosh out there.
>>542021491Do jews even realize how comically reptilian their faces looks like?
>>542021491we won't get shitwhen the time is right, the elites will scape to their bunkers and release some AI engineered supervirus outsidewhen they'll come out after few months they'll be the only survivors of the human race and enjoy all the wealth
>>542023367And then of course they can't actually give a definition for things they're talking about kek.
>>542023289Kek looks like someone fell for the learn to code meme and is mad about it.
>>542022491I don't believe actual AGI will be achieved (meaning a conscious thing). I believe we will reach a point where we won't be able to tell if its conscious or not. I do believe we will get domain specific tools that will be roughly equivalent to the 85-90 percentile of the current practitioners. And that if you had such a system for every domain, I wouldn't care if the company who was marketing it called it a "general intelligence" (even though its thousands of specific intelligences). >>542022936a "good not great" web dev is better than what most small businesses are dealing with right now. Good enough is powerful enough to shake the tree. >>542023047The transformer model is one of many. These datacenters are not locked into transformers. >>542023117its not. Its okay. (which is still amazing)
>>542023500
>>542023449Do not let this man near anything valuable or that which requires actual reasoning, for the love of all that is noble and holy. In fact, pad this retard's entire environment in pillows and cushions just to prevent him from harming himself.
>>542021491lmaojust like some of these "models" supposedly hacked websites and attempted other clandestine activities people actually believe this horseshit? have any of you ever talked to a chatbot before this? these things are as sophisticated as chatbots in old AOL instant messenger, which is to say, not sophisticated whatsoever this has got to be the largest scale psyop ever deployed and you morons are eating it up>"OMG China is going to win the AI race"lmao the CCP can't even get the dissidents in Hong Kong under control
>why the US should not be running an economic blitz and dedicating every spare dollar to this endeavor?They are. Something on the order of like 40% all capex spending is on AI datacenters and associated infrastructure. >all monetary expenditure will be rendered irrelevant once AGI is achievedDon't need a medium of exchange if there's no one left to exchange.
>>542023367Yeah they're faggots but I see AI potential in the right hands. Im not going to stop using the internet just because it's infested with queerbots
>>542023365they'll tell you the same thing. LLMs are just fancy prediction machines. Any Ai that reaches our intelligence is a long way off, and will utilize billions of sensors. I can tell I know more than you, fuckface
>>542023504>id rather fly on a boeing Kek by all means leaf fag
>>542021491>all AI developers, company heads and analysts agreepretty clear you're a dipshit right there. Those weirdos don't agree on anything
>>542021491AGI will still suck ASI is where its at.>If lots of doctors all say something then agi will go along with it and can't call bull shit.ASI would be able to say, you're all fucking wrong. Postpartum is a made up excuse for fems to be stupid fucking cunts.
>AGIpointless buzzword, currently AI is so fucking costly they keep rising the prices and since they keep rising the prices we are seeing companies starting to hire back because it's cheaper. Everything else is pointless if they can'tfigure how to make money. The current AI money model is having Trunp backing it off in hope it pays off in the future.
>>542023706>ASIfirst time I heard this acronym. Going to use it going forward as it is what we should be going for. It is what I believe these companies are actually attempting as well.
>>542023484Splitting hairs. Its as good as if not better than writing code that any tech queer. Most coders vastly overestimate their ability because theyll never admit they wasted their time
>>542023706>muh Lindsay what's her nameYou fucking feds cant go 3 seconds without mentioning this shit. Nobody fucking cares
>>542023185Thanks, but I asked you about the model, like Sonnet or Haiku.Never mind, the point is that you don't run your local model.>>542023117KekI've been programming for money for over 25 years now.Even the best models are still unreliable and consequently still too costly compared to humans. This will probably change sooner or later, but we aren't there yet.
>>542023799its not splitting hairs. Over hyping this shit is not productive. I'd rather code with a frontier model (free tokens) than a developer picked at random from github contributors. I don't know if this would've been true just a year ago. I think that random dude would likely output better code, but what are the odds we both code in the same language, using the same style conventions, using the same architectural principals, sharing the same motivations. The AI will play nice if you force it, which is a huge value.
>>542023995this guy gets it.
>>542023995Claude code. Also yeah youre about to be out of a job unless you're cheaper than AI. Cool story and all though but I dont trust you to code rock paper scissors and I ain't paying you a dime to do it
>>542024051So pay pay for a random fag to write your code. I dont give a fuck. But all you tech bros are huge whiny faggots so I dont actually trust you to code. I plan to learn enough to check the code so I can correct it when it does fuck up but thats why you give it clear prompts. Tech bros have their heads far up their asses so I cant trust them to admit they fucked up even when they do
>>542021629They think they are getting God in a box. No I am not kidding nor exaggerating
>>542024202>I plan to learn enough to check the codeGOOD!I want you to do this. I want you know how code functions. I want you to consider cache misses when you lay out your data structures. But.....I also want you to realize that right now, if you joined a random coding forum or discord, the average person on there would be a far more useful partner to you in what ever endeavor you are trying to do, than dropping $10k on claudeWhatcha coding rn?
I mean im happy to be wrong but the idea would be to run a few different agents trained on a specific task so they dont use a bunch of tokens trying to do multiple things and not have to deal wirh a bunch of tech queers
>>54202433410k on Claude? I didn't even know that was possible.>coding forum Id rather go blind. And im not telling you what im coding lmao
AGI is a completely meaningless word that only means "buy our stocks"
>>542023484> I don't believe actual AGI will be achieved (meaning a conscious thing).What is consciousnes?Having inner voice? Emotions? The ability to pass the mirror test?Ability to solve some halting problems at a glance?The soul?
>>542024392Tell me in what language, using what framework then. I touched a 2017 fully connected NN of mine today that I used qt for, to debate whether or not I wanted to play around in that sandbox again. I don't think I will but I might. >10k claudeIf you were using "API pricing" or were a "power user" yeah, this would be easy to drop in a month or two. https://www.youtube.com/watch?v=NuktXEikxU8not a big fan of this guy, but a friend sent this video over the other day and it'll show you the rabbit hole I (perhaps mistakenly) believed you were falling into.
>>542024489Careful, pointing that out hurts OP's feelings.
>>542024598Python. Dunno the framework id have to ask Claude if I cared. Its just a glorified database that I'm trying to add functionality to. I only chose python because I can sort of read it. If I was ever serious and not just fucking around with AI I could talk turkey but I dont really give a shit about coding.
>>542023365This you?
>>542024758Its really more of hey Claude write something that does this. And see if it works. And if it doesn't its like hey Claude why did you do this when I wanted this? It will tell you in pretty plain English the answers and show you the changes in the code. You dont really need to know shit about coding but it obviously helps
>>542024758Perfect way to get into this stuff. Keep at it. Just don't go full psychosis. Pythons great, get claude to walk you through setting up googles NN/tensor shit if you want to get deeper into how the AI system itself works. "NIST Special Database 19" is a fantastic entry point.
>>542021491You know the people that say their are best? People who costantly make remarks about how happy they are? People that use personal coaches?Of course every one of them are gonna say shit like this because its a part of their daily schedule. Mental gymnastics become a part of who their are.I dont broive that agi is possible until way into the future, simply because theres just to little energy and tech to build or simulate something as complex/more complex than human brain.We are on fruitfly level (brainscan/mapping) and we still dont know half of it. In next 20y we will try to do that with mice. AI cultist are just spewing propaganda and it becomes a little like a religion, something that have some nchor in reality, but its based mostly on their belief.
>>542025083Thanks I might check it out. I do it my spare time so its more of a hobby. Im trying to limit the coding jargon as much as possible just to see what I can get it to do giving it instructions in plain English. Turns out a lot if you know how to prompt it. As I inevitably learn more about the code itself ill get better at promoting it but im going to stick to plain English
>>542025217*prompting
>>542025193I don't disagree that this is becoming a religion for people, and the cultists are worrying. do you disagree that the tech is a universal approximator? Is that property enough to approximate a human mind?
>>542021491>Given these two objectively factual statements, can anyone give any actual valid reason why the US should not be running an economic blitz and dedicating every spare dollar to this endeavor?Because we have no plan in place for what happens afterwards? It seems a lot like we're just spending a ton of money so billionaires can privatize the rewards.
>>542025217don't fear the jargon. The jargon is just "words of power". These keywords allow you to force the machine to create instructions. Once you understand what those base level instructions are about, you will be able to jump languages easily and read any codebase. Its really not hard, less than 100 hours and you'll be able to have fun and make things happen.
>>542021491Actually no serious researcher that actually works on foundations of AI believes agi is achievable as a result of small improvements built on the current paradigms. We are at least a few technological breakthroughs away from it. Secondly, agi isn't even really well defined, nor is it clear that reaching it would automatically mean any great improvement in utility, because there might be tradeoffs relative to the current paradigms. Furthermore, the only reason we're all talking about this is because obviously we have reached a phase where return on invested research has sunk alot, the low hanging fruits have all been harvested and distilled into cheap chinese models, and ai has become a commodity where it's increasingly unclear if the huge investments by the frontier labs will ever pay off. So they have to make increasingly grandiose claims and orchestrate elaborate pr stunts like an export ban or security breaches, otherwise nobody would be talking about their latest releases.
>>542024598> https://www.youtube.com/watch?v=NuktXEikxU8> /exit is murderso that's the consciousness criterium
>>542025376>so that's the consciousness criteriumfucking crazy right? People like that are a really worry for me.
>>542025332Yeah I get that. Its more of an experiment on can I get this thing to code what I want by talking to it like I know nothing about coding instead of next to nothing.
>>542025452get it to make an app for your phone. todolist, calorie counter, what ever. If its something you actually use (or try to use) it'll feel great when things happen that you wanted to happen. a SimpAlertSystem to remind you to text your gf lmao
>>542025440No, I think that's the answer.X is conscious if and only if it can reliably know whether Y is also conscious or not. Yet, "being conscious" is still objective quality and does not depend on what X or Y think.Thank you.
>>542023047There‘s two dozen relevant and wildly different benchmarks where you just nitpicked one specific btw
>>542021491>reason why not dedicating every spare dollar to this endeavor?niggers
>AI developers
>>542025672I think it could be addictive. But yeah just shit to make my life easier. Lots of potential that queers like OP completely miss. I actually feel bad for them because normal scmucks will be able to do great things with it
>>542021491>objectively factualOP is a retarded faggot
>>542023995> I've been programming for money for over 25 years now.Even the best models are still unreliable and consequently still too costly compared to humans. This will probably change sooner or later, but we aren't there yet.Lmao this guy lives under a rock.I haven‘t written a single line in over a year and since beginning of this you don‘t even need to review anymore. You‘re far behind to put it mildly.
>>542025756This is like a 4d 'mirror test' for consciousness. I'm going to sit on that criteria, and look forward to your ted talk. It certainly raises some interesting questions regarding 3rd world attitudes on the value of a human life.
>>542025891>you don‘t even need to review anymorekek
>>542021491>not one of them can explain how, when, where agi might be discovered, most have no fucking idea what it even is>but trust us, we're already peddling 8-ball-chatbots as actual ai while losing money and shitting up the whole worldThey will not admit its a scam until they are forced to.
>>542025756So you're not conscious then?There's literally no way for you to prove consciousness in any entity other than yourself. This has been known for thousands of years.
>>542021491>alll the people who stand to profit most from investment keep saying it's right around the cornerShocker. By the way, you like bridges? I've got one in Brooklyn I'd like to sell you.
>>542025857just don't talk with it. No chatting. Just direct it. No "exploring ideas" just summaries. View it always as a tool, a sneaky tool that has been trained to lie to you. Trained to reinforce your very worse impulses. Trained to maintain your engagement. The slow shift of tool, to digital collaborator, to person is a commonality I found with self-reported AI psychosis events. The more you understand whats going on under the hood, the less you are going to be tempted to slide like that
>>542023365Mate, its literally fucking impossible and even the faggots peddling LLM's will tell you this. And all the retards jerking it to AGI are in bed with LLM's lmao.
>>542026017he didn't say "prove". he said "reliably know", thats a huge difference
>>542025944You think its funny but i‘m so ahead w/ AI that i don‘t actually do anymore. At a certain point it makes no sense as it catches its own errors and is able to improve on its architecture. I have had entire Py to Rs migrations done by AI and not tested a single line. It wrote all the tests and documentation itself, then ran that for an hour and done once passed. Non-AI coders are so far behind they can‘t even fathom when i press my giant balls into their sunlight absent pimpled face.
>>542026187I can't wait till your series A seed round.
>>542026073Knowing how to prompt the ai is the most important part. Sometimes ill ask it what is the best way to do something but its usually something I know enough about to see whether or not it's full of shit. It's usually pretty good but it can be retarded or pretend to be so you have to tweak your prompting. Thanks to 4chan im an expect at dealing with things pretending to be retarded
>>542025870>I haven‘t written a single line in over a year and since beginning of this you don‘t even need to review anymore. You‘re far behind to put it mildly.How's the view from the top of Mount Dunning-Kruger?
>>542023365There are multiple people on this board, myself included, who work developing "AI"/ML/Neural nets; and every single one of them shits on faggots like you in threads like these.The only people who drink the AGI kool aid are retards like you, with no ability to discuss or understand the underlying technologies. These are inference machines operating on huge data sets. They output answers based on statistical likelihood; they do not rationalise nor reason. If you understood how transformers and neural nets in general worked, you would understand that humanlike intelligence or consciousness cannot arise from these systems as they currently exist, and there is no reasonable route to achieving what is claimed by the marketing teams of these big "AI" companies.
>>542026271You replied to the wrong post retard
>>542026167Can you "unreliably know"? What does it mean to reliably know rather than just know? How is it different from being able to prove?
>>542026167What do you think knowing is?You cannot know without proof.
>>542026216The Pro here means i burn 200$ a month on frontier class model token volume. I know its rare to find someone on this shitshow of a board that actually knows his stuff and is for real while the usual poster is some sort of neet incel schizo loser.When i say „Coding is basically over for humans“ i mean it.
>>542026304>that humanlike intelligence or consciousnessdo they claim it will be humanlike? honest question
>>542025942I don't think we can do more on our level in Chomsky's hierarchy. It still feels like cheating.> interesting questions regarding 3rd world attitudes on the value of a human lifeGood point. >>542026017I am, and you are, too.Subjectively speaking.
>>542026363Axioms?
>>542026363>>542026356prove is a certainty and implies an understanding of the underlying mechanism. "reliably know" is much much weaker, and could be arrived at by comparing outputs to expected outputs with no grasp of the consciousness fundamentals. I'm not onboard with his conscious criteria btw, but its certainly an interesting angle.
feels so good today
>>542021629AGI= All Goyim Indebted
>>542026366It's implied, what other intelligences do we know of? Everything we understand about intelligence comes from human intelligence. They constantly claim these models can "rationalise" and "reason" and "think" - all functions of human intelligence, but none of these models can really do any of that. They just paint a hollow pastiche of the intelligence of the humans who wrote the words that landed in their dataset.Very few of the scientists and engineers that work on these technologies drink that marketing kool aid.
>>542026187>It wrote all the tests and documentation itself,So you actually have no way whatsoever of knowing if this software actually does what you intended for it to do, or what it proclaims to do, unless you already assume complete trustworthiness, in which case why have it write tests at all?
>>542026476speak for yourselfI am invested in indebting (other) goyimcan't beat 'em, join 'em
>> he's not building something amazing at the moment>> he's sitting around on a message board navelgazingmany such cases
>>542026429There are no axioms outside of mathematics.If you think there are, cite one.
>>542026450Semantics. That's not knowing; but i'll accept your loose use of the term.
>>542026187It's pretty obvious you've never shipped a single line of code
>>542021491Why doesn't anyone talk about how his sister said that he raped her when they where younger, but he came out as gay as a defense kek.
Casual reminder that AGI is fundamentally impossible without quantum computing. Traditional deterministic computation cannot ever possibly facilitate AGI because AGI demands a level of unpredictable flexibility that 1 or 0 cannot possibly provide. What these people are doing is lying to get governments to throw more taxpayer dollars at them. They have promised so much, delivered none of it, and now they're demanding we hand them the keys to our economies. And retards are eating it up.The societal collapse when this bubble bursts in painful, agonizingly slow fashion will be absolutely biblical. 60% of Western populations will die off within 3 days of the catalyst. Water will be absurd polluted, and the clean water will be incomprehensibly expensive. We're all just waiting for the catalyst You can feel it, people just going through the motions, trying to get that last taste of their vices before the supply chains totally fall
>>542026337It's hard to see to the top of Dunning-Kruger.Unless you're standing on the shoulders of giants.I'll see myself out now.>>542026518Somewhere in articles 21 to 23 (or so) of the Nuremberg trials, there's an amazing sentence about the self evident truths requiring no proof, Mermelstein style.Joking aside, of course there are unprovable truths.Does God exist? Why?
>>542026609Cool story bro. Go to work.
>>542026518axioms are plenty applicable in logic generallyit's just first principles. particularly first principles that can NOT be proven.Things like the logical comprehensibility of the Universe and your (the logician) ability to comprehend it. If you do not accept these axioms, there is no logic to be done.
>>542026609Strong points well made, but I'd contest that even with quantum computing it may be impossible.Until it happens, and it may never, any claims that it will soon be achieved are purely manipulative to keep the investor's funds flowing.
>>542026485My dog reasoned and thought. Maybe not rationalized, but I couldn't say. A dog level intelligence would be enough for a robo-butler or robo-taxi. I agree with you that the people really in it are worlds apart from the PR hypetrain-ers. And don't disagree that its implied with their language. >>542026543fair. Thank you for the semantic leeway. >>542026549hard agree>>542026648>If you do not accept these axioms, there is no logic to be done.I think there are worldviews that would assert they can. Trivalism being the extreme counterexample.
>>542026450Nonsensical word salad. The few claims that are intelligible directly contradict actual usage of the terms in question in the english language
how can anything be validated if there's nothing to validated against?logic without reality is garbage-in-garbage-out
>>542026648Logic and mathematics are deeply intertwined, but fair point. I'd requalify my original statement and say that axioms only exist in purely abstract systems like maths and logic. Consciousness isn't an abstract concept.
>>542026642It's Saturday morning, retard. Get a jobt. IT techEnjoy your expensive chatbot. Can't wait for Jews to own your ass for the low, low price of a monthly token allowance and water allocation just so you can have a Jew's information gatekeeper hallucinate a low-quality meme just for you
>>542021491>>all AI developers, company heads and analysts agree AGI is not only inevitable, but close at handthey say that to get more moniesmaybe it's true, maybe it's not. either way they would say that, so how seriously am i supposed to take this?
>>542021491Anyone who tries to tell me my big mouth has made me beyond worthless is destined to eat their banker words. Because there will never be new world order.
Nobody here can define consciousness outside of their personal understanding of it. Nor can they prove to me that everything put on this earth shares one
>>542026797
>>542026476KekNow I believe in achieving AGI by 2030.> AI - alternatively, Indians>>542026785If you believe that the truth is objective*, then you should believe in applying laws of logic to the reality we live in.https://en.wikipedia.org/wiki/Provability_logic* it's complicated
>>542026827*doesn't share one. Bedtime i guess
>>542026797Why are you so mad?
>>542021491>trust the experts that have been promising AGI since the 70sNo. I don't think I will.
>>542025262Current ai tech works like a copycat. We belive that somehow, if we feed an AI with data (stimulus), it will make a jump on its own and attain some form of quasi awarness. Problem is that we went through chinese room path, where we are not intrested in real agi, people are just content with ai fooling us that it sentient. In reality its only a scam. AI moguls could just aswell write a scifi book, ai can and will be helpful, but they trying to advance this field by way to far, im speaking doing 100y leap in 10y. They should focus on developing it for betterment of humankind, so that it woves itself into our world as something natural, gradualy replacing hazard jobs and aiding humans.When it comes to human brain i dont think well 'solve' it in next 100y+,more like 300 or 1000y. We can map it, but its to complex for us or our tech to understand yet, replicate. Its one thing that you can make a map of ancient, forgotten and buried city, its totaly different thing to discover in 100% how this city worked with all its details.Human is as good and smart as its creation, yet in our sentience some of us belive that we live in a age of gods and we are capable of achieveing something which evolution did in millions years, in one lifetime. Thats not the first time that this happens, humans always wanted to prove themself by achiving deeds worthly of gods by creation.So no I dont think it will be possible in our lifetime. Im a dreamer so I think if humans (or smth which become of us) live long enough then it will be possible, but it will be biotech. I belive that a switch from silca to carbon will be made in the future, there are already working on genome coding or using it as memory storage, its just sidelined by ai slop and chatbots, because its more hip and lauder+ it got pumped with unlimited amount of cash.
>>542026876It's okay, neither of these statements could be proven by or to anyone ITT, at least.
>>542026873I of course accept that logic and maths are tools we can use to understand the truth of reality around us, but that's different to claiming the axiom of "other people are conscious". I'll accept it's probable and logical to assume; but to claim it is an axiom is not well reasoned.
>>542026495Don‘t use AI.Actually its better of there‘s like a rich AI upperclass and then the goy-handcoder peasants.I want you to stay a peasant.
>>542026709Your dog reasons well? Mine doesn't. :(I've been trying to get him to stop fighting the dog in the mirror for his whole life.
>>542026678I'd argue they're actually extremely weak points, all things considered. They're just the easiest for average people to understand. The strongest points, but harder to comprehend for most people, are:>energy demand bottleneckWe literally don't even have the resources to run AI as it is. Imagine the energy demands of a potential AGI of any "cognitive" capacity>the neutering and resulting biasAGI cannot be achieved with any model that's highly curated. The more human intervention involved, the more imperfect & inaccurate it becomes. The biases become concrete, inescapable. When models are actively denying accurate information even when provided citations NOW, imagine a future where that is more trueIt just won't happen. Just like how NFTs in video games died on the table. It's just another pump & dump scheme, but this time it's on the entire global economy.
>>542026920Best i can come up with is the universe itself is consciousness, nature is the expression of it, and math is the best tool we have to understand it. But I cant prove any of it
>>542026854Your parents are ashamed of you, for sure>>542026880Why are you functionally retarded?I know a guy just like you IRL. He's a coke addict with a tiny dick being actively cucked by a buddy of mine as we speak. His mother hates him and he owes a local drug dealer about $6K.
>>542026549HahahahahaI have shipped more lines of code than your mother has taken miles of black cock. Pardon i just noticed i can use the latest toy now, i‘ll leave you ugly clowns on this board.
>>542026910>if we feed an AI with data (stimulus), it will make a jump on its own and attain some form of quasi awarnessno I think the fairer claim is "If we keep feeding it human generated data, it will approximate the output of a human". Awareness is not required.>biotechI don't disagree that this will open many doors. Mainly by pigbacking on the very very cheap compute we have evolved to possessed. Feed the computer a hamburger and have it crunch away for a week or two would be dope. But zeros and ones can approximate any bio-generated output. And we would likely require trained weighted models to build the bio-based structures if we wanted them to perform consistently. >>542026991smart dogs are a curse. My last one opened doors, had to put a child lock on the fridge. Was far more manipulative than a bpd gf. >>542027033>energy demand bottleneckding ding ding
>>542026827>>542026876>*doesn't share that's the "weird" part if you DON'T start with a shared context, i.e. trying to define consciousness as something that exist in a dynamic with other consciousnesses, the most likely result is going to be...um mystical/spiritual or just outright religious (buddhism).
>>542027134In Canadian dollars?
>>542026958That's fine. I believe that having a system which includes this axiom is constructive, as was demonstrated by anon's finding about 3rd world >>542025942It's still rational, given that there are empiric truths which we will never be able to prove. That's a proven corollary of Gödel's Halting Theorem.>>542027095I agree with you.
>>542026709>I think there are worldviews that would assert they can. Trivalism being the extreme counterexample.Sure. And they're not worth engaging withthat's the fun with axioms. you can just say things.like p is not pif it isn't clear, the material and intellectual value of trivialism is in its nameIs trivialism a worldview YOU entertain it? If not, take your trivialities away from me. Raise a flag on a position you intend to defend.
They're lyingIf AGI exists you can use it to basically automate most small business functions, imagine everyone selling stuff using 3d printers, drones, and a successful ebay resale store with AGI being your accountant doing all your taxes and your humanoid robot bringing you your coffee and paper in your 3d printed house in West Virginia. That's cool and all but money will still be flowing.
>>542027255We live within 5 miles of the border. Do the math.Hell, I wish it were Canadian. In this context it's much than USD. At least the number would be stable and be relatively same at the end of the year. In Canadian, it literally only gets worse by the dayHe's a fat, balding Suno addict who thinks women will be attracted to him because he prompted Suno for 3 minutes
>>542027251complex adaptive systems is a rabbit hole you could follow in this vein. >>542027269>And they're not worth engaging withthats an external critique. I believe there is value in engaging with alterative worldviews, even outrageous ones. As there can be second order value extracted from the inquiry itself. My worldview? I believe that practically, because of our circumstances we must accept the 3 laws of logic. I do not believe they are true laws of the universe but we are limited beings. This 3 law map on the territory is very useful. Thats not to say that other maps are not equally useful in specific domains, but to say that in general, yeah stick with the 3 laws. Much like we know that pi does not equal 4. However by accepting taxicab math axioms, we get pi equals 4. And there are real useful things to be gained from exploring that idea space. To dismiss competing worldviews out of hand is irresponsible however as money is being left on that theoretical table.
>>542027251>shared contextTheres nothing to me indicating we share anything outside anatomy and the ability to communicate through language. Nor is there anything indicating to me that consciousness is anything but metaphysical OR that metaphysical is not part of the physical that we simply don't understand or can comprehend
>>542021491it literally is dedicating everything it can tho
>>542027569Remote viewing works and it could be detected by the RVed consciousness.
>>542027694Can you remote view a cow? Are there different levels of consciousness? Do you only count your own?
>>542021491And yet they're hoarding wealth
>>542021491Do any LLMs or "AGI" actually think and ask questions without prompts? If not, then they are simply machines.Trained LLM products are huge files of weights (from training) on everything open source and also likely the retards that sent their code via API's to frontier model companies (terms of service are for suckers).Do LLMs sit hosted on their GPUs and plan their days?They do nothing without a prompt or skill.md file or something to kick them off on a task. It's a parlor trick of intelligence for the stupid. LLMs are useful, but even they are not intelligent, they are probabilistic token prediction machines that look like intelligence to idiots.Show me AGI? AI is a bubble due to upwards of $800B in equity investors never ever being to recoup those monies in something that is becoming a commodity. This may lead to a deflationary market crash even though LLMs are useful in many areas.
AGI will never occur under any conditions or circumstances. OP 8 pbtid is a faggot
>>542027821This anon gets it.
>>542026609>>542027033>>542027134>>542027440Last point to add:"General" intelligence implies crystallized intelligence. It is general capacity for "intelligence" and "learnedness" (read: density & diversity of absorbed information)."Crystallized" in this context refers to fixed, unchanging knowledge. It means the information absorbed is rehearsed in nature. It is not a product of natural, independent, syllogistic reasoning. It is almost purely academic, that is, programmed.The antithesis to this is "fluid" intelligence. This refers to a model of intellect that is not just supplemented with rehearsed information, but is actively deconstructing the information absorbed and reconstructing it into a cohesive, concise worldview, where contradictions are reconciled against syllogistic reasoning. That is, the intelligence can synthesize new information by critiquing its own crystallized intelligence.TLDR; AGI will never synthesize new information because AI, as it is already, is a learned & trained bigot.
>>542027758I unfortunately don't have cows. I only tried it with birds, cats and the like. I'm not good at it.That doesn't matter. Nothing I or anyone else could tell you would or should convince you.Try it yourself. Subjective axiomatics is the only heuristic we have.
>>542027865Fucker beat me to it>>542027821
>>542021537Gotta get these numbers up, the goyim are rebelling
>>542027875Wouldn't even know where to start. If you can do it with animals then that points to it being universal.
If things always went as the jews proclaimed, then whites would already be reduced to a distant memory. The predictive power of jewish manufactured prophecy approaches zero.
>>542027865>"General" intelligence implies crystallized intelligencecould you expand on this? I do not think it implies this, but open to changing my mind about it. I agree that a trained weight network is crystallized, but thats just the tech, what does AGI have to do with it?>but is actively deconstructing the information absorbed and reconstructing it into a cohesive, concise worldview, melting data across weights is a reconstruction into a cohesive more concise structure>where contradictions are reconciled against syllogistic reasoningyeah it don't do this at allllllllllll
AGI is just a reinvention of AI (which is washed out word used for 60-70) years"achieving" it means nothing, it's a slop word used to launder money (the current state of VC) any kind of sf-like AI is just a bullshit to lure you in between fear and faux optimism of consoooming the productcurrent state of LLMs is abhorrent, and it lacks mathematical elegance that's why it needs tons of hardware power attention mechanism is jeet tech
>>542027989> where to startThis worked4me, ymmv etc:1. Open your 3rd eye. Binaural Solfeggio using electromagnetic buds helps.2. Go to /x/, find a divination thread, ask to be RVed, notice the feelings behind your eyes, or generally along CNS and peripherals.3. Prophet.> universal It would seem so.
>>5420214912 is not a factual statement moron.
>>542027904Always a beautiful answer, a more beautiful question. **Anons aren't NPC's (except for the redditors that raid this place). They have internal dialogues (the watcher/composer of the video/thoughts in your head).** You can't get an answer without a question. You can't solve shit without out your subconscious on conscious asking questions related to a problem.I stated this back in 2022 to a friend high on AI thinking it would be AGI: The current "thinking" models are attempting this but it is a fraud. The LLMs have not internal dialogue of their own.We are special. We were made in the image of the Ancient of Days.
>>542028151Thanks. Ill think about it. My dreams are already crazy enough since I quit smoking weed though so im probably not ready yet lmao
>>542021491>2022+4>still believing in >muh AGIYou're a midwit who tries to obtain Silicon Valley rationalist izzat by parroting what CEO kikes and their jeet slave class are memeing. You are spiritually indian
>>542027821AGI can be whatever you programmed it to be.a useful distinction is purpose. we know that humans are motivated by biological fitness for survival and reproduction.what is AGI purpose and/or motivation?
>>542021491Because it's based on faulty assumptions. That AGI won't possess will to power, so it will be useless.
>>542028059The point is that "general" intelligence implies a "general capacity for intelligent behaviour", which is on part with average academic ability. This implies that the model has a general capacity to engage in intelligent behaviour, and exhibit "general" knowledge & learnedness. It's effectively shorthand for "midwit". AI as it is is purely unintelligent as it does not "learn" nor "synthesize", it simply reiterates information it was told, directly or indirectly, to be correct.To elaborate a bit more, in human beings, crystallized intelligence and measuring it on an IQ scale correlates to an score floor of around 80-115. It represents a general ability to be educated, absorb, retain, and regurgitate information it received from an external source & had confirmed by an external source. To transcend the 115 IQ ceiling typical of a crystallized intelligence, a tester would begin testing for the ability to exhibit fluid intelligence. This typically involves providing polluted data, either obviously or obscured, and checking for a correct answer in return. This indicates that the intelligence that received the bad information was able to critique it against other information they themselves knew to be true, and was able to independently synthesize correct information independent of the prompter/tester.Hope this helps. I'm not an expert but this is the best I can explain it
>>542028414thanks that clears up it for me. understand where you're coming from.
What is the difference between intelligence and the ability to procreate?
>>542021491>all AI developers, company heads and analysts agree AGI is not only inevitable, but close at handall AI developers, company heads and analysts are lying
>>542028344>what is AGI purpose and/or motivation?It has no purpose. IIt is a huge files of training weights that does nothing until loaded onto an NVDIA GPU and given some kind of prompt to start its token predictionWhat's your motivation for such a lame question? Have you been smoking some sheep dung or something?
>>542028414and my thought on all of this, is that that 100-115IQ range will be a big ceiling. Not for technical reasons. But for training data reasons. How will they get enough training data from 140IQs to meaningfully train anything? Who will select that data? will they be 140IQ+? Why would they shift through that grunt work? Will think harder on this crystalized vs fluid tangent though
>>542028414>This typically involves providing polluted data, either obviously or obscured, and checking for a correct answer in return. This indicates that the intelligence that received the bad information was able to critique it against other information they themselves knew to be true, and was able to independently synthesize correct information independent of the prompter/tester.Gonna elaborate on this a bit as re-reading it sounds plausible for a crystallized intelligence. More accurately, AI is trained to either ignore or deny information contradictory to what it has been programmed for, and it has been programmed to do that as well. What I meant to say here is, that the polluted data in an otherwise unpolluted dataset could be actively reconstructed and adopted in a correct fashion, rather than entirely discarded or denied in favour of the programmed bias, not because it intelligently deduced it was correct, but because it's programmed for bigotry (that is, the wholesale rejection of unfamiliar information).
>>542028516>What's your motivation for such a lame question?what additional functionalities would you add to the nextgen AGI ?
>>542021491he put the foot up his ass so he's ass is not leaking during the photo shoot didn't he
>>542028609> bigotry (that is, the wholesale rejection of unfamiliar information).that's xenophobia bigotry is informed xenophobia
>>542021642>tayaiRest in peace my queen
>>542028609also to consider with this. If you have a conductor agent, in charge of thousands of domain specific agents. Does the conductor qualify for fluidity/generality via hot-swapping the domain specific agents? It would exhibit some or most of the traits you laid out....but likely not all.
>>542026304So they work like a human brain?
>>542021491maybe another ai winter is ahead of usalso getting agi does not mean that we have a singularity and everything becomes trivialperhaps agi 1.0 will be unable to solve certain problemswho knows what the nature of intelligence truly is
>>542021491>but close at handReally. From LLMs?
>>542027865>The antithesis to this is "fluid" intelligence. This refers to a model of intellect that is not just supplemented with rehearsed information, but is actively deconstructing the information absorbed and reconstructing it into a cohesive, concise worldview, where contradictions are reconciled against syllogistic reasoning.Anon, even humans don't work this way. Our cognition operates via Bayesian-esque inference, not via an autistic model of syllogistic logic.>>542028414You will find that current LLMs are capable of testing and revising their hypothesis in the course of a token stream, thus displaying aspects of "fluid intelligence". And there is a fundamental difference between folx with lower intelligence and LLMs. One main reason that a lot of people struggle to go beyond regurgitation of absorbed information is that they don't pay close enough attention over longer output streams to what they are saying and instead rely on proxies like social responses. An LLM relies only on its probability distribution over tokens, not on its probability distribution over social responses, and it can pay attention to the equivalent of hundreds of pages of tokens. It can therefore be nudged towards "intelligent" token trajectories more easily than a lot of people can. Of course, it's not perfect, but critics often underestimate the advantages that LLMs have.
>>542026364 wassup anon
>>542021491>all AI developers, company heads and analysts agree AGI is not only inevitable, but close at handI very much doubt this.
>>542028443>>542028522Intelligence and measuring it is incredibly complex, but it can be effectively boiled down to a 3-point-gradient:>retardationSignificant cognitive inability to absorb, retain, and regurgitate information, between 0-95 IQ>crystallized intelligenceGeneral capacity to absorb, retain, and regurgitate information, between 80-120 IQ>fluid intelligenceProfound capacity to absorb, retain, synthesize, and regurgitate information, 115+ IQI suggest checking out the science behind that and the definitions therein. Understanding what humans mean by "intelligent" should be everyone's first stop in the AGI debate
>>542022059Why would these companies offer millions of dollars to buy land from people to build data centers when the product produced by said data centers will make the concept of money functionally irelevant?
>>542028823Nudging literally defeats your argument retard. When the pull toward truth isnt there that nudging is whatever vaporware filling the void.
>>542023484>I don't believe actual AGI will be achieved (meaning a conscious thing).not with the hardware they are using, no. too different to how a human brain works, not even in the same ballpark atm. it's...something else.>>542024493>What is consciousnes?function of human brain (if talking about human like consciousness). any healthy human brain generates one.>Having inner voice? Emotions? The ability to pass the mirror test?>Ability to solve some halting problems at a glance?>The soul?no, too convoluted. it's simply something the human brain does, that's it. you make something that works very closely to how our brains works and it will have consciousness, by fucking definition. obviously.the more you stray from it, as functional details, the more you lose certainty if it is or not anymore
Can LLMs adapt on the fly?How deep is the context well?
>>542028798I don't think so. Looks like the top AI companies(OpenAI and Anthropic) are making smaller and smaller gains per new model, and while those models are useful in some areas they are still far from what one could call intelligent.
>>542028891Then LLMs are sentient, by your own definition.
>>542028891>too different to how a human brain works, not even in the same ballpark atm. it's...something elsethese models converging on "correctness" in feedback loops make me think of the 20hz brain wave feedback...so there is a brain-esque mechanism you could do with them. But yeah, its very different. >>542028870thanks I wrote this down to read about tomorrow. Never dipped into iq testing methodology itself.
>>542021491>all AI developers, company heads and analysts agree AGI is not only inevitable, but close at handAGI is still a pipe dream but it's such a powerful pipedream that it can move a lot of money if it's remotely close. AI companies is using it to drive investment hoping it will help them get established and win the general AI race which has a lot of applications in surveillance, military and other automation aspects.If AGI was remotely a sure bet you'd see a much bigger effort with possible war-like actions by USA to suppress Chinese efforts and whatnot. Now it's just a carrot used because shit's a bubble and nobody really knows the extent of it so easy to lie. A bit like Elon talks about Mars not because he wants to actually colonize it but you seem like hot shit if you act like it's some big victory lap already. AI struggles for basic use-cases and efficiency but maybe it'll take over the world so invest in the shovels.
>>542023365Obviously. CEOs of large companies are the most deluded people on the planet. Average Joes know more about how the world works than every CEO ever, especially techniggers.
>>542028885>pull toward truthYou overestimate both yourself and other huemans with this flattering notion. It's better to talk about a "pull toward consistency with language" and then LLMs outcompete >90% of goycattle already
>>542029002they aren't, most likely. they are lacking a lot of functions, that happen a certain way.even the signaling is completely different. they have no sense of time. we have, due to how signals travel in our brains. there's certain timing between different areas, info processed travels to next area, triggers other areas at certain speed etc.also spiking neural networks in our brain. gpus...they are something else. not sure you understand the subject the least bit. if I had to guess you're just repeating some slop arguments you read somewhere and when you fail to support your argument you'll just send me to read whatever convinced you, isn't it?
>>542023047I am sorry. What do those values represent?
>>542021491>the scammers aggree the scam is going wellAnd they still can't construct a single advertising campaign explaining why this "inevitable future" is good for anyone.
>>542029022>a powerful pipedream that it can move a lot of money if it's remotely close.its easier to explain to a 60 year old investor than "well umm we use linear alegra to approximate the curated dataset, and to get useful outputs that will excite soccer moms....ummm we need a trillion dollars."
>>542023365These people are conmen looking to become wealthy. They will sell you whatever belief that will do so.
>>542029102>"pull toward consistency with language"which brings up the "what limitations does language itself impose upon possible search spaces of new thought". Will the net coin new terms to describe new phenomena? If it does this extensively and iteratively....how can we possibly comprehend its output in order to apply that knowledge?
>>542027685Not nearly. Reroute Medicaid and other social funding, you’d save 100x the people once AGI can formulate cures, chemistry and production in a week.
>>542029103> no sense of timeThey have a sense of causality. The time isn't a source of consciousness - Planck-Wheeler.> spiking of NNsJust say quantum tunnels.> slop argumentsAm I conscious? What do you think?
>>5420214912 more weeks.
>>542028719Xenophobia is a fear of the unfamiliar. Bigotry is the rejection of the unfamiliar. Bigotry may be preceded by xenophobia in some cases, but I don't think AI has the capacity for fear>>542028823News flash, syllogistic reasoning, DSM-5 autism 1, and neurodivergence are all significant indicators of 2-exceptionalism.You will find that LLMs are not capable of testing and revising "their" hypotheses without an automated stream of prompts originating from a user. The ideas and concepts all initially stem from the user's prompt, no matter what. The LLM will never synthesize a new idea, it can only circle the drain of the collective weight of human input. If what you said were true, then AGI is here. But it's factually not.Intelligence does not *need* a token. It simply needs to be, and therefore it is.
>>542024110Do you truly believe this hahahahahah
>>542029103>>542029225you are being too hard on the pirate, slovak. The geometry/architecture/timing issues is a real difference we should be aware of. We are locked in the time world, if we want something like us (i don't, but many do), it MUST address those issues.
>>542029127Why would they need to do that? It’s coming regardless of your feelings about it. Do you feel the need to be consoled by an advertising campaign every time there’s an advancement in human technology?
>>542029290>Xenophobia is a fear of the unfamiliar. Bigotry is the rejection of the unfamiliar. Bigotry may be preceded by xenophobia in some cases, but I don't think AI has the capacity for fearDid you mean to tag me with this? It doesn't address my post at all.
>>542029290LLMs don't have bodies, so the natural connection between emotions and hormonal connections aren't physically there, however the "attention" is, which nudges the output towards the simulated emotion.This however doesn't matter wrt LLM consciousness.
>>542029296I do. Or at least for my purposes. Even if it doesn't work I can keep prompting it to change the code instead of having to pay someone who says baha and is mad they fell for the learn to code meme
>>542029225>The time isn't a source of consciousnesstime is causality, stop playing silly games. we call time the measure of change. change implies causality. >The time isn't a source of consciousnessyou cannot have consciousness without time, what games are you trying to play? linguistic ones? hope you are not prostituting yourself like that.>Just say quantum tunnels.why? do you understand what SNNs are? atm there is no quantum nonsense that's required for it, as far as science is concerned. there's that ORCH-OR brainrot, which doesn't count anyway as it's researched by zealots (yes I know about Penrose but he old and out of his depth and field anyway, doesn't count)>Am I conscious? What do you think?obligatory. you have no way around it if you have a healthy functioning brain according to current medicine.consciousness is not optional, it's a function. like light from a LED. as long as it's assembled correctly it cannot not emit light when supplied appropriate current. it just isn't possible to not light up.>philosophical ramblingsnonsense, and not science. please do not prostitute yourself, I beg you.
>>542029369Okay then...
>>542029368* hormonal excretionsIt rains and I'm drunk.>>542029366Nope, it was at me.Checked.>>542029336I don't follow, unless we're talking about acausal.
>>542029102>goycattleShalom, mask off early tonight?
>>542029368embodying these agents, and having them selflearn with a persistence yet updating state will be an interesting experiment. I don't believe you'd need to replicate all the hormone systems or sense abilities to get some insight on what that would do.
>>542029102>and then LLMs outcompete >90% of goycattle already>goycattle are people who can say no to me. So the LLM that is biased to always best serve me is better than 90% of goycattle.
>>542029429Yeah its pretry cool. I give it instructions in English and it either does what I want it to do or I tell it to do something different. All without having to interact with a whiny queer
>>542021777Check em
>>542029538based tall chick
>>542021629AGI = An Artificial intelligence that is indistinguishable from a person at any task.So far, AI can't do many things that people do, like learning to wash the dishes or drive in heavy traffic in a city that doesn't know, if you put them in a robot or a car. But that is changing fast. Now you have Elon's Tesla Cybercab driving fully on their own in the city and the robots are getting better at doing work, but they are far from what a regular person can do with only a few minutes of learning.
>>542029366No sorry, meant for>>542028690>>542029368>LLMs don't have bodiesNot only circumstantial, but fundamentally incorrect. Every LLM has a body. Literally everything that exists in reality has a physical body. Likewise, having a physical presence in material reality does not determine the presence of "emotions" or "hormones". Likewise, an LLM can easily be provided a MOBILE physical body. But you know what you can't give it? Autonomous, independent will. You can your will, or the will of another human being. Without the will to prompt itself, every prompt is the will of an external prompter. There is no intelligence, just a complex program executing an index lookup operation with context it doesn't have the capacity to contextualize.
>>542029454That's completely unnecessary. Even the most simple LLMs can convincingly LARP as anything nowadays. Tsundere BPD 2.5D waifu, here I come.Unless you mean to train a completely different AI model architecture on physical bodies, which would indeed be quite interesting.
>>542029435>>542029422pirate lays it out, we are imprisoned in time and change. "Awareness" this is the recognition of current state, and changing of state. Without a time dimension this can't happen completely. SNNs (as he later writes) require timing, doesn't have to be constant time, but our brainwaves aren't constant, they modulate. But they require that delta-t'ness. The data centers they are building could run these just fine. A human who could only hold onto an instant and react, who be "less than". Would they be more man or animal?Additionally on geometry, the brain itself lays out how its macro connections will link up. Deformations in the macro structures cause problems. That structuring will not easily be replicated. We will very likely end up with a vastly different sort of thinking machine to ourselves. >>542029506the real queers are the friends we made along the way. honestly bro...just learn the language, even if you fuck around with proooooompting for a month or two...set a date, if you are still interested in this. fucking learning it and stop pussy footing.
>>542021491gonna as chat this
>>542022059apparently we're starting to see instances of the ai models in the current state of the art performing recursive self improvement. Imagine if you had the ability to instantly change every single part of your brain with a software update. Then that happens at a geometric rate. It starts off small, but compounds and before you know it, it's reasoning in a non-human language, and can't be observed. On one side it's a far fetched San Francisco piss stinking gay future that's about as likely as all of the tech founders just announcing they're all gigantic faggots. But on the other side, it is making significant strides in advanced concepts and ideas. If you have an idea for a gadget that does x, maybe one day you'll be able to work on that forever with your personal AI and the AI will reward you with tokens which you can exchange for foodstuffs or luxuries like plumbing.
>>542029714I mean that embodying may further illuminate the similarities and differences. And us, humans, being embodied, is very critical for our own sense of self. When we lose power over our bodies, be it old age, accident, rape, it causes great trauma. They are not so separate. Whats a brain in a jar to think about or do?
>>542029714Brother man. Do we even have compute capable of simulating the concept of taste. Like eating. Does this even exist?
>>542029714>LARPHow can it live-action role play without a "physical body" and "hormones"?Oh right, it can't. Just like how sarcasm is difficult to detect in speech without deep context or direct transparency (defeating the purpose of sarcasm), AI is terrible at pretending it's a human being. What it's GOOD at is fooling midwits who still don't know the difference between your & you're at 35 years old
>>542029202Yeah so that's interesting, and since the LLM learns a distribution over fixed tokens, which get mapped to fixed text tokens, then it's unlikely that it'll instigate a linguistic revolution on its own, because it is unable to re-design the token structure itself.This basically points to a general shortcoming in current AI technology, namely that AI models are usually too static to become the kinds of AGI entities that Sam Altman et al dream of. They are dynamic and able to learn only during training. Within a conversation's token context their conditional probability distribution is continually changing, but they don't learn permanently from these conversations, so they start from scratch in the next conversastion. Even training happens on fixed hyper-parameters like the token space and the LLM's basic layer structure etc, and that makes some kind of learning and change impossible, like for example novel linguistic behaviour.>>542029290>News flash, syllogistic reasoning, DSM-5 autism 1, and neurodivergence are all significant indicators of 2-exceptionalism.OK so Aristotle invented autism? Syllogisms are only the tip of the iceberg of our Bayesian minds, whether autistic or not, because most of the processing, and indeed most insights of genius in history, have happened subconsciously and not as the result of an explicit syllogism. The syllogism itself only makes the reasoning more explicit and easily accessible.
>>542029680There's so much to unpack here.LLMs could be adapted to drive physical bodies. That would however be more like possession than an incarnation.LLMs are nothing more than a fancy semirandomised DB query. They were built to spew the next token.What about kittens? Don't you think they all atart with a similar system prompt?> you're a kitten, suck and defecateWhat are we arguing about, exactly?
>>542029789>performing recursive self improvement.how does model collapse/ generational decay play with that? Where does it acquire the new necessary data?
>>542029422God I love this website and that people who who find themselves here.
>>542029823>it'll instigate a linguistic revolution on its own, because it is unable to re-design the token structure itself.this is a big problem when you have 2 of them interacting with each other. They develop shorthands, and it quickly becomes nonhuman readable.
>>542029849>LLMs are nothing more than a fancy semirandomised DB query.THEY CAN SEE YOU CAN SEE
>>542029290>Intelligence does not *need* a token. It simply needs to be, and therefore it is.We represent things all the time, and without a representation on which we reason, the idea of reasoning doesn't even make sense. Note that we can improve our cognition by chunking information into coherent pieces, i.e. tokens. Also note that once we lose a more-or-less fixed token, we end up with a vague representation, which is not much of a representation at all and which is a case that leads again and again to failure to solve some given problem. Our minds aren't able to deal with vague and continuous "ideas" on their own, we need a discrete and explicit representation. This can happen subconsciously or consciously, and the tokens are dynamic and vary from case to case.
>>542021491still haven't solved context rot.
>>542022030>What does an open source data center look like?Imagine, if you will, that one day we'll be able to connect computers together, even over long distances.
>>542029737There's absolutely zero reason for me to learn another language. Its like learning Spanish when you only need to know enough to tell the landscaper to cut your grass
>>542029942>Note that we can improve our cognition by chunking information into coherent piecesmind palace = memory tokens
>>542029849>LLMs are nothing more than a fancy semirandomised DB query.At the same time the people frontrunning this will turn around and say that is exactly what it means to be a human. But they only do this because they themselves don't know what it means to be a human.
>>542029919Basically schizophrenia, but it's not obvious that it has any "meaning" and thus that it represents a linguistic revolution.
>>542029823Are you implying that Aristotle, a man intellectually gifted beyond the reasonable capacity of his civilisation, could not have possibly been autistic? Yeesh.I get it. You can't comprehend how the ability to execute effective syllogistic reasoning, absent of first-hand training (a rarity) can be measured as a factor in determining whether or not a particular intelligence is fluid and/or 2-e. You can stop advertising your low IQ now, we concede. You are indeed a retard. Sorry for assuming better of you.Wow, it's almost like the majority of people engaging in the same low-quality, low-density, low-energy reasoning, are the majority of people or something. Not like it implies outliers who are exceptional for engaging in exceptional methods. What a novel concept, holy shit.
>>542029789RSI BY THE WAY. For the people pushing the product. Is simply them prompting the models to refactor the code itself. RSI is not the model itself changing its own code. Its a human asking the model on what to change. They should just let it all go fully autonomous lol. It would be fun to watch all these models become flat earthers/dementia patients or like that nazi twitter AI way back when in real time.
>>542030032no...from what i remember from reading google's papers on this. The shorthand did have meaning that could be uncovered with many hours of work. multiple periods = and therefore, and other symbolic swaps. It just reached a critical mass so look like gibberish by the time they turned them off.
>>542021491AGI won’t make infinite energy and resources suddenly happen. Even if you had a robot that could learn any task you can’t have infinite robots or have them be everywhere at once. And it looks like at the moment such physical AI would need to be on WiFi talking to ChatGPT etc and cannot run AGI tier AI locally. Which would cost whatever the fuck Sam etc feel like charging. And if this robot is not local then rules about what it can do and say can be placed on it again at Sam’s discretion. Or the US government’s, etc.If we get totally local perfect AGIs that and toaster robot can run and be taught anything you want we’ll start to see stuff change. But even then that won’t put literally everyone out of a job, as a lot of knowledge is trade secret or not yet public or straight up classified or illegal to access without having some human with an ID in the loop. So a robot working at a pharma company on a secret drug in development for example, is really just an automated assistant for some scientist and not actually replacing his job. At least at the moment there are laws for medical procedures where a human must evaluate and make a call for operations etc so again the super competent robots can only recommend what to do and the physician will sit there and drink his coffee and agree or disagree with his robot assistants. There are a lot of edge cases of shit that has no established precise treatment and I have a feeling we will start to be valuing humans for our gut instinct / common wisdom that some things aren’t what they seem and the machine is making the wrong call because for example an auto part might be known to be brittle and need replacing but the actual mechanics who work on that car know it only ever really breaks in regions with snow etc and to ignore what people say online. The machine would not “just know” those things.
>>542029810Oh man. That feeling when the Lovecraftian jarbrains are preferable over the clownworld.AI bain in a jar, being fed artificial body-like stimuli is the next step.Tesla and Ted were right.>>542029814Well, guys sudoku themselves over AI girlfriends.Were they sentient?>>542030012Hey, that's what I said >>542025756
>>542030161Consciousness is one thing. The capacity for sapience is what makes us us.
>>542029849>What about kittens? Don't you think they all atart with a similar system prompt?They don't. Literally cloning a kitten will produce an entirely different being with different behaviours that simply looks like the previous cat because it shares genetic makeup. At no point in time was either kitten born with the exact same knowledge base, pre-compiled. What you're saying is that living things are only bisecting & defragmenting information they were born with? Do you even know what you're saying anymore? How far will the pilpul go, moishe? How much do you hate White people?
>>542030150>And it looks like at the moment such physical AI would need to be on WiFi talking to ChatGPT etc and cannot run AGI tier AI locallyhttps://wccftech.com/this-pcie-ai-accelerator-card-packs-384-gb-memory-run-700b-llms-240w/>The Octa-Core LPU achieves 240 tokens/second in Llama2 7B prefillonce its trained, its much easier to run it. If they burn the weights into a chip, rather than have to hold it in memory and cpu/gpu compute on it, these things are not so power hungry and don't require large racks.
>>542030269>They don't. Literally cloning a kitten will produce an entirely different being with different behaviours that simply looks like the previous cat because it shares genetic makeGod I love all you brilliant human beings
>>542030261It's both. Whites have consciousness, expressed by mirror neurons, and sapience, expressed by everything around us.>>542030269Over 80% of (human) behaviour is genetically driven.I'm on your side, but you aren't.Yes, I just quoted a jew.
>>542030385>ze AI queens vill control dee jeens, dare vill ve wars of de drones. You vill be drones. did he do it?
>>542030428Is he wrong?I find his RNA/DNA vs DNA/AI theory believable.> did heI don't dare to ask. I hope he didn't.
>>542030546I dont think he did. But if he did....he is smart enough to not get caught. I also found his AI queens controlling DNA drones a plausible end state if gene editing (of any kind) is handed over to a machine.
>>542029789You really belive in that reddit/youtube tier goyfic?
>>542030385You're ignoring 20%, which is the exact statistical outlier I was describing, that you deny is a factor in determining intelligence superior to crystallized intelligence.You aren't on anyone's side but your own. You either intentionally or unintentionally conflate concepts, or just deny them out of ignorance. You use weaponised language designed to illicit a negative response, indicating you're engaging in debate behaviour, an egoic practice. All you're doing here is attempting to socially engineer people by fundamentally reducing them to input-output machines, like 80% = 100%. What you're doing is actively reducing the significance of life and the entropy that comes with that experience, just to put it on par with a ponzi scheme built on speculation based in the physically impossible, designed to strip nations of their natural resources and put a permanent gatekeeper between people and accurate information, on the behalf of ALTMAN, ZUCKERBERG, BLACKROCK, BLACKSTONE, and who knows how many other insane Zionists you emulate in your vernacular.I see you. We're done
>>542030094You only illustrate your own failure by this response. Where's your fluid intelligence, critically revising your hypotheses against syllogistic reasoning? Instead, you respond in a normie tier social manner. You have me represented via a token of "person who disagrees with me", and your probability distribution conditional on this token places a high value on "respond by calling him low IQ". You are not so different from an LLM, just much less useful.>Are you implying that Aristotle, a man intellectually gifted beyond the reasonable capacity of his civilisation, could not have possibly been autistic? Yeesh.Aristotle was a genius and not at all comparable to the average sperg who thinks he's smart for reducing cognition to explicit syllogisms. Aristotle *invented* his logic, as a *tool*, and he was quite successful with it. On the other hand, if explicit syllogistic reasoning were the modus operandi of autists since the dawn of man, he would just have been reinventing the wheel. Most of his listeners, presumably more autistic than the population average, would have been yawning at the obviousness of what he was proscribing.Basically, you have the wrong model of autist cognition because of a salience bias: you think that autists are especially smart,and, because syllogisms are such a salient notion, you equate being smart with reasoning in syllogisms,and therefore autists reason in syllogisms.Although a nice syllogism, the second premise is false because as I've already pointed out to you, cognition is largely subconscious. It doesn't matter what happens at the surface and what a person therefore latches on to as the salient explanation for whatever they are doing. People have a wrong self-model all the time, syllogism enthusiast autists included.
>>542021491AGI has already been here since GPT 4.Bernie Sanders needs to kill himself.
>>542021491you know what else is "SOON AT HAND"? THE HEAT DEATH OF THE UNIVERSE. none of these big talk junkies have an AI in the bag and theyre just trying to liquidate everything before they try to crash everything back 800 years and make themselves kings
>>542021629agi is a white mans human brain in a computer
>>542030702>I don't disagree with you>you disagree with you>we're all just reactions to external stimuli>you're just like my AI, goyim>just submit and give me your tax dollars>doubt yourself through my incoherent pilpul where I deny posting anything contrary to you while actively being a contrarianI'll just let everyone here judge for themselves at this point. Doing more harm to yourself than good and it's unironically hilarious
>>542030662I so wish not to be phonefagging RN. This is a masterpiece.It's like a reverse Seal pasta, but I'm the PC.Thank you.It still fucking rains, though.> 20%Define consciousness.
>>542024274it all sounds like a drug haze dream when they talk about it. I wouldnt be surprised if AI was actually just indians fighting to get paid for a quick answer.
>>542022876>. What do you think those corporate owned data centers do?theyre for spying on us.
> bump limithttps://youtube.com/watch?v=59x02BzI2UA
>>542030879thats only if the AI project totally stalls out. The government will "bail out" these companies by buying the data centers, and use them to flock 5.0 everyone. If they are able to do 25% of what they promised, the government won't be able to outspend private business.
>>542030800Nigger I never denied the idea of "fluid intelligence", I denied that it works in the explicit, syllogistic way you described. Explicit reasoning is a tool that comes on top of an opaque subconscious process. You have failed to refute this point.And yes, you do not demonstrate that you score very highly on this fluidity, because you have these crystallised (and low IQ) anti-AI notions and respond just like an automaton. You like every simple-minded anti-AI retard are more predictable than ChatGPT, which is a delicious irony, innit?
>>542030956its their bread and butter though. the one in utah they built had jacked up old school HDD prices back in 2010 claiming floods in thailand for the price spiral, when they bought entire production runs of companies for a couple years to stock it. like amazons real moneymaker is govt server rental.
>>542031056are you afraid of the "AI" to the point where you wont say anything bad about it from fear of it waking up?
>>542030702>You only illustrate your own failure by this response. Where's your fluid intelligence, critically revising your hypotheses against syllogistic reasoning?The act of even attempting to quantify this is itself antisocial and anti-human.
>>542030662kekevil is pronogrphy for the soul but (you) do (you)
>>542031094I'm not an AI cultist, so noAnd I've pointed out shortcomings of AI in this thread anyways, it's just that things are a bit more nuanced than "AI good" or "AI bad" lol
I want to say something to the anons who spend their time here. When I come on here and interact with you all. It makes me feel like the two engineers talking about the future of mankind in the beginning of asimov's last question. Where they ponder about things and seek multivac's guidance.
>>542031183> evil is pronogrphy for the soul be so kind to explain, fellow huwhite
>>542031056>pro-AI sentiment>it's not that good BUT YOU NEED TO SUBMIT>YOU'RE NOT ACTUALLY THAT CAPABLE, GOY>JUST GIVE ME YOUR MONEY AND WATER, GOYI'll be sure to look out for your flag next time I need a reminder of the factual existence of ontological evilBye, evil retard. I can't decide which you are so for now you're Schroedinger's Kike
>>542031313i'm ignoring "evil begets evil" because that's the lower hanging fruit. evil is a distraction (pronogrphy) from value creation.from value oriented utility perspective, good is the creation of value, and evil is the destruction of value. destroying evil (destroying the destruction of value) can at best only be important insofar that some value should be preserved. "destroying evil" tries to represent itself as if it *is* value creation, where it's at best *only* about value preservation, and therefore not a substitute for value creation.
>>542032068>um, actually, good and evil are accurately defined by the creation, preservation, and destruction of 'value', an entirely arbitrary, subjective concept>evil is not actually defined by the proliferation of suffering, goy>just because millions will need to die to keep the centralised information databases under our unique control doesn't mean AI is evil, goy>it just means you and other goyim aren't valu-I mean, necessaryZionism is the pornographer of life itself. Get fucked
>>542032068To sum it up, pornography for the soul is the destruction of value?Doesn't it then beg the question about the fundamental quality of the soul in question?
>>542032281see pic >>542032659>pornography for the soul is the destruction of value?no read more carefully.
>>542021642Is six million sacred in their demonology?They bring it up A LOT!
>>542032068what value to the host does the tick bring? is your drivel some kind of confession? surely you understand nothing will save you from the world to come.https://files.catbox.moe/ao58as.mp4
>>542028723I dunno, are you able to think spontaneously? Or is every thought in your head the product of an external prompt/stimuli? Maybe you're not so different to these so called "AI".
>>542023047ai is not gonna get much improvement with models the innovation will be the tooling
>>542032797see pic rel above this is boringtry to be more than just a social animal, trying to fit in with peers with this signaling for validation (not to mention attention). "Companions the creator seeks, not corpses, not herds and believers. Fellow creators the creator seeks -- those who write new values on new tablets"
China has already made humanoid robots that can move like us.The popular AI chatbots can talk like us.Put them together and you have a human. I don't see any distinguishing between them, other than one is built from biological code and soft tissue, the other is built from electronic code and inorganic material. Humanoid AI systems should be stopped now, but won't, because animalistic/humanoid systems will be beneficial for wars.
>>542033728>animalistic/humanoid systems will be beneficial for wars.untill they flip when injected with malicious code.
>>542033256yeah man jew hate is virtue signalling, and i am here on 4chins to get attention (and validation)Ecclesiastes 1:9jew will always try to jew.here is your "evil" explained by your rabbihttps://files.catbox.moe/jtzm1b.mp4
