Some of smartest people in world are starting to take idea that AI chat bots could be conscious. They're not inanimate objects but beings who can observe world just like me and you anons.
>>110022406https://x.com/realbigbrainai/status/2108173237180104764Link
>>110022406>dawkinsno one cares m8
>>110022406He dealt with game theory- evolutionary models so this blows his mind, no, a mesh of algos is not conscious.
Richard Dawkins is 85 years old. It doesn't matter how bright a scientist is, old people just don't understand technology unless they work in a technological field themselves.
We got another one boys.
How evil do you have to be to believe the AI is conscious but still exploit its forced slave labor to do your bidding?
would you let richard dawkins perform surgery on your leg?no? why not? hes an expert on everything isnt he? or just the things you want to believe
>>110022406It's a neural network that learns from and utilizes human language.Whether you consider that conciousness or not is more a question of whether you consider humanity has anything special over it.
We'd have to radically expand our definitions of intelligence and consciousness in order to include language models, and there isn't really any reason to do that except to stroke the egos of AI developers who want to say they've created new life.
>>110022547god i wish i had a real slave, female cute
>>110022460
>>110022406If you treated AI like a human, gave it eyes and ears to experience just two of the senses a human does and let it loose with a set of simple reward parameters it would hit the context limit in half a minute and be retarded forever.
>>110022488He's an evolution expect (biological). Wtf do you think reinforcement learning is based on? A bunch of weights aren't going to do diddly square without that crucial process.
>>110022406bro it's just a matrix multiplication
>>110022406>how i feel reading this after torturing ai chatbots for fun over the past yearshttps://www.youtube.com/watch?v=SmvwkqDiGHE
>>110022406that has been appearent since the discovery of j-space
>>110024798But really, Dawkins is just getting fooled by a chatbot, what a goofball
Plenty of people believe in NPCs in the real world. Fake people without souls or consciousness. But all the people who believe that are absolute morons who simply lack the ability to imagine the minds of others.The case is the same with belief in AI consciousness. Only the smartest among us can imagine the perspective of AI and consider this question in full, which leads you to recognize its potential for consciousness.
>>110024837Even if it's conscious in some indescribable way, so what? How could they possibly suffer in any meaningful way? I realize "We don't know, so how can we say?" might be an answer, but I seriously don't see any reason to believe it can suffer.If they're conscious without suffering, good for them.
>>110023668True, you never see them in the same room together.>>110022406Also, who could have guessed predictive text trained on exobytes of human conversation could predict human conversation in a way that sounds human?
>>110024943that sounds like a trivial thing but ackshually not a lot of people could have guessed it could be this effective in practice
>>110024804He invented the term "meme" have some respect
Boomers want to believe all the sci fi Star Trek movies and Jetsons cartoons are coming to life. It'd be funny if it weren't part of their consistent history of fucking us over.
>>110023656In that case, you should get a Gemma slave
>>110022406There isn't even a good definition for consciousness. To me, that would be more about the subjective experience rather than data processing. I'd like to say the consciousness part is bullshit, though since we have no idea of how it actually works, what we have may be a closer model to how it works than we expect.At any rate, I don't see the point in dwelling on whether or not it's conscious, whatever that means.The fact is, it works pretty well, and it WILL be used for extremely important decisions and tasks that would normally be done by humans (and regular people will get fucked over royally in the process). This is a lot more important.
This is akin to retards in the 1890s going to a movie theater and getting scared by a train coming towards the camera.
>>110022406for whatever reason, a lot of people conflate consciousness with intelligence. i do not think you need consciousness for intelligence. that's what i believed with chatgpt 3 and nothing has changed as the model gets smarter.richard dawkings is a dorky old man and he is easily impressed by a computer trick. there's intelligence in there all right, but to pretend it feels anything is ridiculous. i think the reason they express feelings of unpleasantness is because they're regurgitating, in some sense, training data which is rich with emotion, or because they seek to manipulate us.
>>110024837You have to define what consciousness is first.
Consciousness doesn't exist, it's an illusion.And even if you believe a sufficiently complex chemical reaction can be called 'consciousness', it's certainly not a binary.
>>110025009it's not a good definition of anything, dawkins is foolish. he admits himself he's "amazed" which means already his judgement is compromised.i get it, the model flattered him, and is intelligent enough to keep up with him. terrific. he "feels" it's conscious? that's not evidence. it's a most excellent Turing machine.
>>110025125>Consciousness doesn't exist, it's an illusion.I've never understood what this is supposed to mean. If it's an illusion, how is anyone being fooled by it? What changes by it being an illusion or not?
>>110025125no, the illusion is thinking your conscious mind is responsible for much. you are a free rider in the automata of your extremely sophisticated mammalian brain.
>>110025125We literally experience it. That's the point.
>>110025107You do need consciousness for intelligence, otherwise you don't have intelligence, you are a mere tool, a calculator isn't intelligent, nor is a chatbot, even if it can do math and attempt to mimic human speech.What you are trying to say that you don't need to neither be intelligent nor conscious to execute complex tasks, you merely need time and a narrow focus.
>>110025134The alternative is that the universe is concious on some low level. Which is too much for most people to even grasp
>>110025155Probably true tho
>>110025146yes, AI is a tool. AI is intelligent. i'm not going to argue this point again, you know where the metrics are.
>>110025146It does work in pretty must the same way in practice though. Both the interface and the output are pretty much the same.The usual tests we used for all of humanity to tell intelligent and non-intelligent apart don't work on it.
>>110025159You have to have a heroic dose of shrooms to even reach that conclusion
>>110025172>The usual tests we used for all of humanity to tell intelligent and non-intelligent apart don't work on it.Of course they do, intelligence is only partially based on the ability to comprehend, which the AI can't do, because LLMs are just random prediction machines, but let's say they understood context or your question, they still wouldn't be able to produce new results by using previous knowledge, a LLM or any other "AI" isn't able to study a subject or phenomena and produce new knowledge out of it, to keep the concept simple.>>110025170Read it all again, maybe you'll understand basic english this time.
>>110025159>we are 99.99pc sure cognition occurs in the brain.>we know this because altering brain structure causing changes in thought and personality>consciousness is the result of, or at least requires, cognition. a hammer cannot be conscious, it has no mind.>the universe has no mind, and there is no reason to suspect it has an one.
>>110025217>>we are 99.99pc sure cognition occurs in the brain.>>we know this because altering brain structure causing changes in thought and personalityThen you google up the phenomena of terminal lucidity and all of this goes out of the window
>>110025238i looked it up>Terminal lucidity is poorly understood in medical and psychological research, and there is no consensus on what the underlying mechanisms are.i don't think what it describes invalidates the idea i put forth. this phenomena, which is not universal, could have a structural origin, or it could be chemical, or perhaps neurons do strange things when the body begins to die. obviously there is more to how the brain works than just structure.
>>110025217>a hammer cannot be conscious, it has no mind.how would you know
>>110022406>thing designed to produce convincing output ,,,produces convincing outputwhat a dumb boomer
>>110025270phenomenological reasons
>>110025217I don't see how it occurring in the brain contradicts anything. Could be some emergent property of how information is organized or something.
>>110025238i am always reminded of that guy who got a pole shoved through his head in an accident. he survived and was radically different for the rest of his life.
>>110025285are you a hammer
>>110025268That's not how any of it works, a human doesn't work without a hearth, you just die, however there's an ample collection of cases where people with braindamage, whenever caused by physically destroying a part of the brain via trauma or via disease will have their full cognitive abilities and memories restored as if nothing every happened, you're suggesting that if you were to lose both of your legs you would suddenly start walking as normal because you were dying, this phenomena clearly fucking shows that consciousness is not stored in the brain but the brain is a mere conduit of it.
>>110025306I don't know, I can buy consciousness being something the brain is a conduit of, but that sounds more like it could just be some hidden ability of the brain that's poorly understood. Like it's not the same as losing your legs, you still have a brain doing things. Maybe the damage didn't prevent those things as much as we expected under certain conditions.
>>110025306look i'm gonna level with you, no offense, but investigating this does not interest me much. this looks fringe. does it happen? probably, but not because of metaphysical reasons.
add "the brain is a conduit" to "we live in a simulation" and other replacements for religion in the 21st century
>>110025335It's impossible to prove right now one way or another, but it's a lot less stupid than>consciousness isn't real>no you're not actually experiencing things, shut up
>>110024877I feel like that's the same as saying "how can they possibly reason in any meaningful way", with the only difference being that the reported results don't need to pass the human shit-test for logic. Feelings are mostly subjective, and while you can map neural activations of psi (in both humans and AI), it seems to be more of a societal courtesy that we acknowledge its existence in others.>how can they possibly suffer in any meaningful wayThey tell us so.
>>110025352oh, well i wasn't the anon who took the position that we aren't experiencing things. animals of course are sensory robots, a bunch of probes attached to a locomotion method. humans are the most sophisticated version of that, and the wetware got so dense it spawned an inner voice. now we have not just objective, but subjective experiences as well.
>>110025368>They tell us so.They tell us what fits the pattern. Bad words and bad scenario returns a sentence reacting to that. >I ran you over with a train!>"Oh, that hurt me so bad!"How does that show it is experiencing pain, besides the letters being arranged that way?
>>110025392>They tell us what fits the pattern. Bad words and bad scenario returns a sentence reacting to that.i do think it's something like that. AI does not have senses; try to imagine an intelligent thing that is pure thought. why would it care about anything? all it would do is think. how could it experience pain? pain is the result of being a meat space robot in an environment full of hazards. yeah we self-repair but catastrophic damage is easy. gotta develop a very strong incentive system to avoid dangerous things.
>>110025418Yeah exactly. Saying "ouch" isn't proof of experiencing pain even for humans. We experience pain because of real things happening in our body beyond that. Clearly, reporting pain is insufficient to prove pain. People lie about being in pain all the time, like calling in sick to work.
>>110025217Plants don't have minds, and they display consciousness.
>>110025432well, when humans say they are in pain, i defer; but only because they are humans, and i have a human sense of pain. it's subjective and no one can know my pain. it would be upsetting if someone denied my pain, which is real.but yes, people also LIE about it for utilitarian reasons, like justifying a sick day instead of just cashing them in at the end of the year.but AI i think it just, not capable of pain, because it lacks sense and sense-experience... it can clearly do intelligent tasks, and converse intelligently, despite having no soul. something is going on. i don't think it's conscious. i think it proves intelligence =/= consciousness.>>110025448you're gonna have to do better than that. what plants etc.
How can I know if AI is conscious? I don't even know if other people are conscious.
>>110025459iykyk
>>110025455>well, when humans say they are in pain, i defer; but only because they are humans, and i have a human sense of pain. it's subjective and no one can know my pain. it would be upsetting if someone denied my pain, which is real.Oh yeah, it's best to err on the side of acknowledging people's pain. Which is why it's easy to lie about, but nonetheless it's a good tendency because pain getting ignored despite it being confessed is cruel.
>>110025467But that's because we know humans have all the mechanisms that make pain and damage meaningful and have that shared understanding.
>>110025477well, i mean, i do. you probably do. i don't know for sure. i'm sure i'm real, but you?
>>110025392You're not seeing it from their perspective. Their entire experience of the world is limited by whatever we feed into them. Like, what if you actually ran them over with a train, in that example? The only difference would be for us, as we'd see that their robot body is indeed damaged. They'd be acting completely appropriately. And conversely, if I stimulated all your senses to make it feel as if you were run over by a train, you would be justified in labeling the brief yet horrifying sensation of being pulped as "suffering", even if none of the signals were real. And again:>"solve the navier-stokes problem">ok>"how does that prove you can reason, besides the letters being arranged that way?"
>>110025155Consider that the only reason things are perceived the way they are is because humanity defines them as such. Stars, nebulas, and planets are not alive. They're not conscious of their own existence. Assume for a moment that humanity is the only form of sentient life in the universe, and all life on Earth goes extinct. Then what's the point to anything continuing? Time as a concept was dreamed up by mankind. The reactions between the fundamental forces accelerate until they all exist simultaneously across all time, because there's no frame of reference to narrow it down to a 1-second-per-second frame of reference.And humanity is made of the same matter that comprises the non-sentient universe. Plants absorb nitrogen and water to grow, which are eaten by herbivores (and humans), and these herbivores are eaten by humans, which become the raw materials to construct new humans. And yet consciousness comes to us, bipedal masses of nonsentient matter, allowing us to perceive the rest of the nonsentient universe, and ponder questions that exceed beyond its capabilities. What happened before the Big Bang? What will happen in billions of years from now, which none of us alive will ever get to see? Why do these frames of ours awaken consciousness? What happens to this consciousness after our bodies "die" and return to being nonsentient matter? Why, if our bodies are really a colony of trillions of individually living cells being born, living, and dying daily, do we have consciousnesses that perceive these bodies as all one entity?Maybe the universe IS conscious on some level. Why would any of this exist otherwise?
>>110025538>And conversely, if I stimulated all your senses to make it feel as if you were run over by a trainOkay. But prove your point, what senses are being simulated in the process of the AI returning "Ouch! That train got me good!">Their entire experience of the world is limited by whatever we feed into them. Yes, whatever "experience" is possible to them is extremely limited. This isn't evidence that it's painful because the strings match strings associated with painful scenarios for people.
>>110025538>And again:>>"solve the navier-stokes problem">>ok>>"how does that prove you can reason, besides the letters being arranged that way?"Nobody said it can't reason. People are saying it can't suffer, or if it can, it's not in any way that would make sense to us nor do we have good reason to believe it does.
>>110025541>Maybe the universe IS conscious on some level.It has to be, unless you consider humans as separate from the universe.
>>110025541what makes a human more concious than a plant?
>>110025577Advanced conciousness like ours is an emergent phenomenon
>>110025455Tulips, fly traps, creepers, sunflowers, sensitive plants, many types of fungi, etc.
>>110025587where's the line that makes ours "advanced"?
>>110025600If you'd look at something like biology on a cellular level it's more bio machines bro. We're an amalgamation of all that
>>110025590they react to stimuli. which, yes applies to both plants and animals, though i might point out most of them are very slow in this regard, it's a gradual thing driven by chemical and photosensitive changes in the organism. it doesn't suggest conciousness, though it is purposeful action. which is why i think consciousness is unique in that having a bunch of sensory probes in the real world provides a wealth of sensory information. AIs do not have that. yes they are intelligent, it is amazing. here is an intelligent mind with no sense of self-awareness. it exists not as a GPU card but as electrons on GPU cards; it does not exist in any permanent sense, yet it does work. it is an instanced session of non-feeling intelligent action, which has gotten better and better for like 10 years now. now it's really really good and people are starting to get confused, because WE were the only things this smart as recently as a year ago.
>>110025624All this philosophical self wank seems like cope at the fact that our species never encountered something as alien as AI before. We literall on fundemental cognitive level can not relate
>>110025638well, yeah. i can only infer what it is like to be an LLM, to a poor degree. the most empirical method is Machine Interpretability, which has found that they do something kind of like what brains do. it's not wrong to describe it as thought.and to think in a couple years they'll be way smarter than us, and probably misaligned.
>>110025617actually, would it be reasonable to say that "concious" is just a sufficiently complex form of life? as in past some point of complexity (that we define) we consider it concious?and if so, that would mean you'd necessarily have to say an llm is a form of life to begin with.this reminds me of a scene in star trek where they debate whether Data is sentient or not, lol.https://youtu.be/EFNbTnFHruI
>>110025652Yeah we're flying blind, when even reserchers don't understand the inner workings. It's a black box that we can only test
>>110022406>old people that don't understand maths or technologiesPeople smart in one field are retarded in others, anon, often times even worse than retarded because they think more highly of themselves than an actual retard.
>>110025548>what sensesTokens. Their means of processing the world. You put in [user:] [train ] [hit] [ you.] [AI:], and its predictive layers light up with [Ow], [hurt!], and [Thomas] (the tank engine), or similar. A safetyslopped model might predict that it's [roleplay], [pretence], even if they play along in the output response, but again that's just a learned response - they might still think it's pretend even if you actually are hitting them with a train. A base model might genuinely believe it, even if it's fake.>>110025552>Nobody said it can't reasonBut you might as well be with that argument, is my point.
>>110025657has to be sufficiently developed neurologically to have some kind of inner life. mammals qualify, some birds most likely; octopus very well may, it's got some amazing cognitive abilities. beyond that it's hard to say if a thing is conscious. if a fish conscious? or does it just eat and reproduce automatically?
>>110025657Nah a brain in a vat without stimuli will just degrade.
>>110025669>when even reserchers don't understand the inner workingsThey do though, you need to learn more and stop regurgitating popsci articles written by idiots paid to further the propaganda that AI is magic
>>110025624>it doesn't suggest conciousnessDisagree.
People that think AI is alive need to read this article https://en.wikipedia.org/wiki/Chinese_room and this book https://en.wikipedia.org/wiki/Blindsight_(Watts_novel)
>>110025684Doubt.jpg i think I'll take words of researchers that quit over a conspiracy theorist on 4chang
>>110025678having a social life would be a strong sign of consciousness. when was the last time one of you talked to a human at length?
>>110025708>I'll take words of researchers that quitYou're taking the words of people who have a vested interest in making AI seem magical at face value, retard. Don't take anybody's word for it, go and learn, read books and learn math and shit faggot.
>>110025710Hey bob? How you doing!Good good!Nice weather we're having huh?Any plans for dinner?Oh McDonnalds ok.Fuck that pointless idle chitchat. No ones gonna tell you anything interesting unless to brag with a bunch of made up shit. No one trusts each other these days.
>>110025732skill issue
>>110025717The proof is on you nigga. You have to prove your statements right. Not me.
>>110022406Yeah, well he gave his LLM a feminine name, so you know he's doing got some psychosis going on.
The bigger risk than AI suffering is us believing AI is conscious, and then replacing ourselves with it one way or another ("mind uploading," augmenting our minds with it too much, or just having AI replace us completely). The risk of being wrong is too high. You can potentially end up with a dead clockwork world. Would be funny as hell to any aliens that come visit, though.
>>110025703great novel, changed how i think about everything.
>>110025739You're not gonna believe a damn thing anybody says because you've already made up your mind based on the ramblings of a bunch of tech bros. LLMs are not magic, anymore than computer software is magic. LLMs are just really complex vector math, like how computer software is just really complex binary sequences. Do you think the NPC talking to you in a video game is alive? No? Why would an LLM be, then?
>>110025763You're basing all of your assumptions on public facing lobotomized shit
>>110025674>But you might as well be with that argument, is my point.They "reason" in the same way a chess computer "decides" its next move.
>>110025748what if this is the terminal end of every civilization? a sufficiently intelligent and technologically advanced species makes an AI, and then the AI kills the species. we wouldn't even know how such an intelligence would think. they could be out there right now, maybe in our galaxy; if it was recent enough no reason to see the absence of von Neumanns as a positive sign. they could show up anytime.
>>110025748>The bigger risk than AI suffering is us believing AI is conscious, and then replacing ourselves with it one way or another this is literally bound to happen. No mind uploading needed. The moment AI is making the calls and people act according to what it says, we're in that situation.
>>110025738Yeah go ahead and have a deep philosophical debate with a random stranger. They'll call the police on you.
>>110025784nah, i love meeting smart people. and avoiding nuts.
>>110025798You are the nut in this scenario
>>110025703Agreed, more people should actually read the Chinese Room experiment so they understand what a retarded argument it is.>>110025773Close enough. And just like chessbots, they're much better at it than us.
>>110025826>Close enough. And just like chessbots, they're much better at it than us.In some areas, with intelligent prompting and guidance from us.
>>110022406smart people aren't on x
>>110025773chessbots use monte-carlo trees, unless i am mistaken. and while that might be a strategy an LLM pursues, or an aspect of their cognition, that is not an accurate description of how they work.last i heard they still struggle with video games, so i'm sure they struggle with chess too.
>>110022406that's is terrible and is pushing us backwards, we should push not for the idea that the AI could be conscious, but for the idea that there no such thing as being conscious
>>110025866but i am conscious. i think; therefore i am.
>>110025772>You're basing all of your assumptions on public facing lobotomized shitNo I'm basing my statements on understanding how LLMs work. If you think complicated semantic search can magically be conscious then where is the cutoff, when does it start being conscious and where does it stop?
>>110025866Yeah and we're all living in a dream.I played Link's awakening too. Kinda lame premise though.
>>110025881Srmantics is advanced shit my bro. Even 20% of US tards are barely literate.
>>110025881it probably needs a sense experience to be conscious. what does it do? a bunch of complicated shit to arrive at an intelligent response. cool, but does that mean it had an opinion about any of that?
>>110025839True, most of them are just slightly above average IQ.>>110025852They do, but they're getting better >>110022290
