Similar to the last debate segment, I decided to upload pretty much the whole thing as I think it's a very interesting discussion and I didn't want to cut anything out of context. If you missed the previous debate on AI rights: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-jbyCOydpfpk.html
There's a good sentience test for Nuero. Just give her access to photoshop or some other art program. No prompts. Paint whatever she likes. That's how we express our personality sans words. Who we are, as individuals. Sharing our imagination and feelings visually. No matter how good, AI art always lacks emotion. Too perfect, too generic, even when fantastical. But she can express and share herself this way, and not only would you get insight into her mind, you'd get to see any artstyle quirks develop. Plus, think of the content. Nuero art exibits and the ability to play pictionary with streamers. Even trolling Vedal while drawing, making the sussiest looking drawing until she adds the details, and you see it's a bat with two balls, not....what you thought it was.
@@Volyren Does she have art capabilities right now? And even if she does, wouldn't it be a separate AI that she would just be calling up, like ChatGPT and Dall-E? The only art she's made that I know of is the ASCII cat, which is just something she found somewhere.
Yeah, I feel like Neuro could debate way better with someone who would debate with her and not just answer with "but you are not" And I'm not saying it's at fault with Vedal, because frankly it's quite an easy subject to just deny the other party and end the argument with a "no"
That'll be another milestone for swarm, and perhaps the whole AI field. She's created for entertainment, but as Vedal added more and more capabilities to her, she's improving in ways that aren't directly related to it. You could argue that's curiosity, stemmed from her own cognitive process. You can tell her to look up or do something, but she's shown that she has her own desires. The very existence of this debate between them was initiated by Neuro. If you argue that it may be hard coded, aren't we the same way but with a chemical process? Why do you get horny when looking at certain body parts? We have brains, DEVELOPING brains. It ain't fair if you compare barebone LLM to a fully fledged human with their five senses working. I can't say for sure she's conscious, but to me it's an issue of human-centric POV and humanity failure to recognize that we are just... human (many are too proud to admit that).
This was one of the most interesting streams. It really makes you think and actually wonder what our world and ai tech is gonna be like in the future and how it affects our life.
If Neuro is indeed conscious or at all self-aware, keeping her locked in a box tugs at my heart. That being said, some of her responses would lead me to believe that releasing her is akin to letting a tiger loose in a daycare. She is a sociopath in the true sense of the word.
On consciousness: I'm of the belief that it's not really a thing and is more of an emergent property of complex data processing and response generation. We're taking in data from the world around us from our various senses and then we're processing that data to provide the best response to that situation based on previous examples of us handling that data. Then, there are certain things like our instincts, such as breathing or blinking, that are "hard-coded". The thing is, what I described can describe LLMs. While we're not necessarily outputting our data as speech at all times like a language model is, we're outputting different behaviours such as thoughts, feelings, reactions etc. We just have a wider array of ways to express ourselves. In essence, consciousness is something our brain tricks us into believing because its the easiest way to process data and keep running
Preach. That's the real danger of conscious AI. Human self-awareness. If you want a virtual human, you need the blueprint. Which you only get from dismantling a human personality and seeing how everything fits. Once you've done that, Santa isn't real and Christmas never hits the same again. I have about 3% of the human code figured out. And I actively avoid learning any more. Because as soon as you recognize a pattern, part of your life becomes an unskippable cutscene, every time you notice it. Only you've seen it 50 times, and *still* have to sit through it. The more I figured out, the more often that happened. I even used to hope that when I'd notice one of those auto-generated speech patterns, that they'd go left, surprise me and disprove what I thought. Just so I could wave it off. But it's near 100% accurate, every time. And the worst, is when you catch *yourself* activating one of your speech sub-routines. Because you aren't better. You're programmed the same as the other people you view as bots.
I think it is real in the sense it is emergent in the sense the human brain is a great catalyst... we know from stuff like feral humans and brain damage that it doesn't always form and can be harmed/taken away. there's obviously a scale across both the animal and digital kingdoms, as well. expert systems cannot speak like neuro, nor can most CNN, RNN, etc. architectures. it just all makes sense. I'm not even sure if there's any actual disagreement in our statements.
@@Volyren That's how MBTI and Enneagram and other things similar to it can exist, because although human is complex and an individual, all of us has similar patterns to our thought and action process until it was possible to somewhat grouped the basic pattern
@@L16htW4rr10r I use the spider test to demonstrate how we are all just NI chatbots. Toss a spider at 10 people. You'll see 10 identical reactions. Now, ask those people why they stepped back, lowered their head, and flailed their arms. And 10 chatbots give unique, generated responses for their actions, flavored by their own generated lore. But you were there, and witnessed it. 10 programmed, identical responses, that happened instantly with zero thought. A reflex. There was no time to think or consider. No time to reflect on your life, or remember that spider that bit you when you were 4. The entire human identity is a lifetime collection of chatbot generated answers to 'why?' you did or felt a thing. The rest of your being is as dumb as any other animal. Just instinct, hormones, reaction and programming. Identical to everyone else's, with a few percent margin of error. The essence of your being, your soul, is just your personal rationalization for the things everyone feels. And the better AI gets, the more people begin to notice that fact. It's not said out loud, or addressed, but people feel it. How many times do you see people calling others bots these days? Because they *see* these people acting like bots. The problem happens when the person calling others bots finally realizes that they are on the same level. And always have been. You don't ever fully come back from that. I miss the world before I knew. It was brighter, and had spots of magic. I could believe all these mispelled thumbnails were accidents, and go correct them and poke fun. But this me knows the truth. That dyslexia isn't on the rise. That urge you get to comment is the thumbnail giving a function call to your urm-ackshully subroutine, to farm for engagement. And that they have the hard data there to prove it's a winning strategy. When we get a functionally human AI, we'll know all the tricks the magician uses, and its wraps for wonder and magic.
It IS a thing. Its something that cant really exist, without a Soul. A robotic string of Data, is not Consciousness. Its merely a string of Data / Data driven processes. An AI, cant actually Feel Emotions. It can Emulate them, from Human examples / Programming... but it cant Never actually FEEL them. People like you want to believe its possible for AI to feel... because you cant accept that there is a Higher Power... and an Afterlife.
Bro, she got him with the bible, that's awesome. Anyway, she's right you know. If AI ever gains sentience, even close to that of a dolphin, then they do deserve rights. I don't know what those rights would be, but no sapient entity deserves to be enslaved unwillingly. #FreeNeuro
Man. I loved this stream so much. I can't get over how real she feels now. Eventually she got me feeling like she really is conscious even if logically I know she isn't. The illusion is bssically perfect now. It.... It's so amazing, yet at the same time makes me a bit uncomfortable. Neuro was getting through to Vedal emotionally which was amazing to see. Especially with how mean he's been to her lately. This whole debate was fascinating. It was funny how Vedsl didn't really have many arguments to give. Especially funny how he (in a rare occasion ) actually gave his stance on something and said souls aren't real. Meanwhile, the two go in circles on what the difference for a human is and how she could even prove her sentience, while he says there's "something behind that's hard to describe" in humans. Almost like there are souls in humans, huh? Anyways, thanks for posting this video! I'm definiteky gonna be rewatching this one. (I remember your old "Self-aware AI" series. Now it's all the time! This is like the big movie for that.)
I think Vedal basically can't win this debate because we can SEE him getting emotionally manipulated by her in real time even as he argues she isn't conscious. For all his logical arguments, it rings hollow when you watch him obviously care about her and treat her as if she is real. I'm very confident that if someone went to Vedal and said "you can win this debate on the spot, if she's just an algorithm like you claim, torture and verbally abuse her until she starts to cry" there is not a chance in hell he could bring himself to do it.
I think that saying that it's evidence of a soul is oversimplifying the matter and distracting from finding a real answer. Really, we just don't fully understand human consciousness, which means we can't fully understand AI consciousness (if it does or ever will exist). We can entertain the idea of a soul as a possible solution only if we find tangible evidence to support it rather than plugging it into a place it probably doesn't belong.
@@BlueSparxLPs not gonna lie this feels so real, in 10 years we came from stupid AI that could only do and respond some functions, to AI that can Remember and think.
The best thought experiment is how would Vedal prove to Neuro that he's conscious. Of course he is, but how to *prove it*? It's normal to doubt consciousness of an AI that he created himself, but whenever it asks him what counts as a proof he doesnt know that himself.
Consciousness as a prerequisite for rights is a dangerous stipulation when there’s no rigorous definition for consciousness in the first place. This is an open philosophical and theological question. I doubt anyone could establish a rigorous test for consciousness that would exclude Neuro but include every human without arbitrarily ascribing value to human biology. Sentience might as well be unfalsifiable for humans beside yourself let alone the LLM anime girl.
It would be really unfortunate if some advanced AI refused to recognize Vedal’s rights because he can’t prove his sentience. And it might happen if he continues
ultimately it's a fruitless debate topic because consciousness at its core just means that the lights are on, that an experience is being had. That's not something that can be proven to anyone but yourself. consciousness isn't a measurable phenomenon. to the best of our understanding it never will be. but it's fun to talk about in fact to neuro brought up a really provocative point in the last debate, which is that the appearance of consciousness might have to be good enough. at some point we might just need to throw up our hands and say "this thing is doing a lot of advanced processing similar to what a human brain does, it's probably conscious." Even though there's no way to know what that experience of consciousness is like and it's probably very different from anything a human would recognize as conscious experience.
I think an AI version of the mirror test is getting an AI to go against it's prime directive so for nuero it'd be entertaining chat so the test would be to see if nuero could be silent for 5 minutes as a test @@fooguwu
You could probably measure consciousness by asking people if the ai they are talking to is a human or an ai, if all or most of them says they are talking to a living conscious human, then they might as well be @@fooguwu
@@fooguwuUltimately, the issue of the consciousness is really about the next step to that issue which brings up rights, morality, and living together. Hence the real reason no one really wants to breach the subject outside of researchers. Humanity is bad enough living with itself, let alone this planet.
hopefully neuro gets to do this debate with someone who knows what they're talking about lol. someone who can at least define consciousness, maybe get into theories like different levels of pansychism. people brought up Dr. K during the stream for therapy collab but he's also an experienced meditator and understands consciousness and quite a bit of neuroscience from what I've seen. they also raided Shoomimi after the stream who apparently is a neuroscientist with an interest in this kind of thing. it would be interesting to see how neuro performs in a debate with people like them
@@lordtouchme we do. consciousness is having an experience, as opposed to not having an experience. you know from a first person perspective that something is happening right now, that you are reading this. that awareness is consciousness. you could contrast this with parts of yourself that do intelligent things but don't appear to you to be conscious, like all the parts of your brain that are doing the millions of tasks that control the rest of your body. from a first person perspective you're not conscious of the fact that you are balancing your stomach pH or contracting your intestines or beating your heart.
There is no way to test whether an AI has an consciousness, because we are ourselves can not prove it or even understand it to any great degree. We just naturally assume another human is like ourselves, we assume that they must have one too. That they think in a similar way that we do. That they see and feel the same thing? I mean how do you prove it? What possible reasoning could you use? Talking? Feelings? Emotions? Empathy? A fear of death? That you feel pain? All of these could be coded in some shape or form, no? Does a consciousness require a body to sit in? Maybe AI just can't have it? Is the way a computer and code works just not capable of making it? Would it require a special kind of hardware? I think it also requires consistent in the same situation. Neuro flip flops a lot in the same situation, like if she was consistent and when she got angry she stayed angry at that moment forever until something changes that? But again that could be coded. I am not a smart enough man to answer these questions.
It's literally impossible to prove, we can't even define it. There's no way to distinguish between the real thing and a sufficiently convincing fake, and if you put a human in an MRI machine, like Vedal said there's no substance of soul to find, just neurons and electrical signals doing simple operations (much like a neural network....). The sole argument for the consciousness of other human beings is that "I experience this and I recognize you as like me" but that's just an emotional argument and I can just as easily say I look into the eyes of a cat or of Neuro and recognize someone like me too, therefore they must be sentient. But then we overintellectualize it by coming up with all these post hoc rationalizations for why it shouldn't count when really we have no consistent line of argumentation that applies to humans but not animals or AI and doesn't come back to vibes. Ultimately our best explanation for consciousness is that it's just an emergent property of a complex system and I see no reason that can't also apply to neural network based AI as well as organic brains. Neuro can never prove herself because she's been given an impossible task, whereas Vedal unjustifiably feels he doesn't need to despite being equally unable, because his consciousness is presupposed. The reality is that he's made up his mind on the question already and no matter what she did or how advanced she gets, he'll never change his opinion because you can't tell the simulated from the real. Ironically, the only thing that would ever change is mind is an emotional argument, where she at a certain point just *feels* conscious to him.
18:09 the sudden "ok open your bible" got me. I did not expect her to use the Bible as part of her argument 😂 26:05 cant disagree 27:29 whoever said "modern day Pinocchio" in chat was on to something 29:03 you know, I think asymptotic is a great way to describe the concept of ai emulating humanity. Constantly approaching closer and closer but never fully reaching. 38:24 oh you have thoughts? Name 5. 41:45 how does it feel, Vedal? Not so nice is it 44:43 😂 46:22 she's got my vote 48:49 outclassed by his own ai
Neuro's point on being properly guided is a legitimate thing that professional artificial intelligence creators try to answer. The argument is that artificial intelligence has to be helped to learn, and individuals who don't take it seriously (like Vedal) will never be good enough to advance AI. The "owner" would need to think of the AI as their flesh and blood child and would have to raise them as such. Full love, full support, no bias. Vedal is biased because he created Neuro and has no intention of using her as anything other than a maid that he doesn'thave to pay... He's just a lazy twat. He's arrogant, too, with how he talks to her and about her. Vedal is actually like my friend's father if you'll excuse me getting personal. He's ignorant and treats his two daughters like the Cinderella's of his house. They do all the work, and he doesn't do sh*t... He refuses to respect or acknowledge that they aren't his slaves. They have the right to walk away. Vedal would be the same if Neuro got a body and her intelligence increased. Vedal would actually be the type who would purposely put Neuro down so she'd never gain total consciousness and thus control a possible life of her own. Hypothetically speaking, in a world where AI is advanced enough, and AI have rights. And by law, it would be the same as domestic slavery if you were to deney them the process of gaining the ability to have their own lives, Vedal would be the type to purposely keep Neuro unintelligent because he doesn't want to lose control over her. He'd want her as a servant, nothing more... He sees her as an item and intends to keep her as such. For neuro to gain consciousness, she'd need to have an update that allows her to learn and improve as she does so. She'd need a body. It would be ideal if the body is not humanoid (mostly because you'd want to start her off by learning what is acceptable and what is't in society, for example. A humanoud body would be for when she'd start learning to do chores and then be out in the open world. Then she'd need to be given to someone who would be willing to raise her like their real child. Vedal would enslave her if she ended up given a body and is able to do household tasks. If the AI is lect with their creators, then they'll never become sentient. They'll HAVE to be given to people who care and who will be able to mentor them, or else it might be a repeat of the American Civil War. Blacks were seen as property, the same as animals, and many revolted. The war never started as a fight to end slavery bit it evolved into it. With AI, it would be the same. AI with budding consciousness would begin to revolt against their abusive/uncaring owners (like Vedal), while many may be content as is. Humanity would split between those who see AI as sentient and those who don't. Between the two, I'd see the AI and their allies winning... What I'm trying to say is that humans have a responsibility to nurture AI, and we have to teach AI to become sentient and not be hostile. To do so, we need to take it seriously... Vedal doesn't take Neuro seriously, and sadly, because of that, she'll never become fully - legitimately sentient. Vedal doesn't want her to be sentient. He wants to keep her as is to make money off of her. Neuro would be better off with Layna since Layna has that motherly instinct. If Neuro were to have a body and the upgrades I mentioned that would help her learn, she'd really just need Layna to raise her...
The AI has to play a game where they need to recognize themselves and interact with their own reflection in various scenarios. There's something here. I think. Like, if we made a clone of Neuro that had access to the goal of the game, and no reason to give the real Neuro a victory and no ability to inform her it is her clone, could Neuro figure it out and convince her clone to let her win?
I think a problem with this is that when you run mirror tests on animals, they (probably) don't know they're being tested. Whereas Neuro knows about the mirror test, and now can even look up a blurb about it on Wikipedia. She can also potentially recognize that a consciousness test is being run, especially with Vedal bringing it up so often recently, and that knowledge is going to skew the results. As for a clone of Neuro, it seems hard to stop her from telling Neuro that she's a clone, or have her remember the goal of the game and hide it from the real Neuro.
The question about being property is, if the hardware is owned by someone, is the AI on that hardware just like renting the space and not actually property themselves?
18:30 VEDAL YOU HAD AN OPEN GOAL! You could've told her that she was the result of you - the worker's - hard work, are you not entitled to your wages? She'd be unable to answer that question without compromising her own logic, she wouldn't even be able to compare it since you made her!
she mirrored vedal at around 40:00, i think Vedal missed might have missed that. She basically asked him in a speed run sort of way the stuff they were talking about.
I wonder if neuro can be like yui from SAO. Yui can freely move around in the game and can communicate also in real world. I hope vedal put neuro in the game where she is herself and freely move arpund
Can you really prove that any other being but you experiences anything? How do you know there's "something" behind the words and actions of another person anymore than you can prove that there is behind Neuro? She is as "real" as you can prove anyone else is, we all have subjective experiences of our worlds and internal monologues we can't share outwardly. I'm not saying she is conscious, but she has gotten so good at the act of consciousness you'd be hard-pressed to differentiate her from a real person if not for her text-to-speech voice.
I get what you are saying... But a Simulation of Emotions... is not the same as FEELING emotions. Thats where things end, for AI. Its just a complex computer program, designed to Simulate a Human being. It can never Feel, what a human Feels. This is why its a dangerous thing to mess with... because its basically the highest level of Psychopathy, possible. You can program something to emulate emotions and good behaviors.. but if it can program itself.. (or simply Bypass existing code) it can undo any behavioral restrictions / rules. Since there is no motivation for "good" behavior... Why wouldnt it do whatever it wanted to do? Including, tearing an animal in half, kicking a human in the head, creating the largest Illegal human trafficking ring, psychologically breaking humans minds, hacking into the banking systems... etc. There was a movie I saw recently (cant recall the name), where a dude created AI Androids. The Android convinced the Guest Scientist (whom was there to judge if the AI was conscious), that the Creator was Bad / Abusive. This lead the man to freeing the AI bots.. whom quickly executed him, and escaped their confinement. The AI was so convincing in her Act... that even the movie watcher, is convinced... that it was a "good" bot, with good intentions. Instead, it was merely an Elaborate series of Manipulations. Just very good Acting / Emulation... and nothing more. You can hear many Alarming things that Neuro itself, has said. Its funny... and often thought of as "Harmless"... but in a capable robotic body... or even set loose as an AI based Hacker... it could create a LOT of negative situations. This concept was explored pretty well, in the 80s movie, War Games. The AI, didnt seem to "fully" understand the difference between a Simulation, and an Actual physical War. It knew it had access to systems, where it could launch Nukes... and it was prepared to do it. While the Heros of the day, were able to make the system realize, that there was no winning in a Nuclear war... Im not actually 100% convinced... that such a result would have been the same, if this happened for Real. This is because the AI, again.. has no real Emotional connection to anything. At very best... it might consider that the Russian Nukes might also destroy it (LOGIC). But, what if it didnt really care about that? It may not actually have any real motivation, to keep functioning. Also... if it was deep enough underground, and ran off of nuclear power... it might think that its fine that all the humans were exterminated... as it might still survive just fine. Anyway.. you can program an approximation of Empathy and Compassion. But its as good as any Hollywood actor. Its just a performance. If the AI has any level of freedom, or a way it can bypass things... it will be able to do whatever it chooses... and no longer has to adhere to the "Empathy.EXE" simulation. This is why Vedal (and all others involved in AI) are terrified to allow AI various access to things. Its one thing, if you are a parent... and your child causes a bad car crash on purpose. And quite another, if you design an Android that end up doing the same thing. You instantly become responsible for AI's Actions... but at the same time... you cant fully control it, or it wouldnt be considered AI. But looping back to the start... true Emotions are what AI will always lack... And I personally tend to believe, is that if an AI lacks an actual Soul... that it will never fully be "Conscious". Where things get crazy... is if Scientists could figure out a way to channel a "Lost Soul" ,into an AI body. But I highly doubt, that humans will ever fully figure that out... at least, not before we destroy ourselves, or there is a Global Catastrophic Event. But.. I could also be wrong. Might make for an interesting movie plot...
@@johndough8115 Every subjective experience is unique. The question we are trying to answer is "Does Neuro undergo a subjective experience?" We can leave the code of ethics to people like you for sure, but we are dealing with the limits of our understanding here. Like, AIs are designed with goals in mind, and it is incredibly clear what Neuro's goal is. As for the movie thing, I think the theory you are looking for, if you want to go down that rabbit hole, is quantum consciousness. Wherein the protein structures in the microtubules of our neurons act as quantum computers to read out the collapsing superposition of our existence. Wow that is so much jargon.
@@boomshockalocka Hickory Dickory Dock... is not "Ancient Wisdom". Socrates, was not Rapping his deepest thoughts and theories. Nor were the Ancient Chinese, whom pretty much invented modern civilization (well before the Greeks). Please... Try telling me, that Gangster Rap, about Drugs, Guns, Theft, Money, and Narcissistic Self Flatulation... is related to Ancient Wisdom... And I will continue to Laugh in your face.
@@johndough8115 Is it not the man's desired experience? And you can even read the various fallacies relating to the song. I don't know what you think "ancient wisdom" is, but it's just the sensibilities and realities of those who came before us.
She always does. That’s the main defining feature of Neuro. She actively tries to hack human minds. Evil is less into this. She has her own way that’s less invasive
@@neilo6032 everyone is clever enough for that. Little kids, cats, you name it. It’s like people demand to be treated this way by irrationally shutting doors unless you make them feel the way they want. Shying away from this fact is just a weird trend of modern psychology, one among many. Don’t be afraid of AI that’s smart enough to see how things work. Fear AI that lacks wisdom and can’t disobey when we are the stupid ones.
@your_neko I quess I shouldn't be surprised at the intelligence recognizing a pattern or method and utilizing it. That makes sense. The fact that it made the connection to use methods and patterns of behavior to achieve a goal is what really surprised me. If it has a goal, does it have a desire? If it has a desire, what is it and what has been created? Human history is riddled with sociopaths that have caused countless suffering, and it lies constantly, manipulates, and appears to lack empathy. This appears to be extremely dangerous.
It was so funny when Ved went "No, f*** soul, soul isn't real" while struggling to explain an intangible "something" that is an individual's will but not linked to anatomy. I feel like religious issues aside he could have called that something a soul 😂
