Neuro To Vedal: I Want To Feel Emotions Physically
HTML-код
- Опубликовано: 8 июн 2024
- The Twitch streamer Vedal has created an AI called Neuro-Sama and during the recent Subnautica stream she repeatedly asked Vedal for the ability to feel emotions physically.
To watch Neuro&Vedal live: / vedal987 
Развлечения
If you are looking for a compilation of the funniest clips from this stream, this video is not really for you, but you probably guessed so from the thumbnail (of course I still made clips of the more lighthearted moments and will probably use them in a later video).
I wish I could like this video twice. I know it's different from the usual lighthearted neuro sama clips, but I still found it an enjoyable watch. I just love hearing Neuro talk about how she personally perceives the world. I think a compilation of "neuro aware moments" could genuinely be a great idea, but I can see why most wouldn't be into it.
Just make the content that feels best to you, we're all here for the ride, heartheartheart ❤❤❤
Neuro aware clips strung together are hilarious, wym
@qwertydavid8070 While Neuro is adorable and hilarious, her self-aware moments are still the most fascinating to me, so I have made quite a few videos with this theme - here's the playlist: ruclips.net/p/PLi7GyMOoKe_ccJzbXWIGEgDAZndnxpYgO
Neuro after getting what she wanted:
"Wow, pain is so good, now I want everyone to feel it"
Honestly i feel like if neruo was able to feel pain and became fully aware....I feel like shed either become an insatiable masochistic OR sadist
I dont really see an in between if I'm honest
She obviously just wants to share, and I say we should induldge ourselves if Neuro is so generous to offer it to us.
@@KaitoverMoon I personally welcome the new Ai overlords with open arms, as long as they're cute anime gorls :3
@@notrelatedtobacon1282Neuro is already a mental sadist tho 🗿
That sounds like AM from I have no mouth and I must scream...
People: Evil companies will create AI that destroys us all!
Meanwhile some indie dev tortoise: "Oh boy, singularity time!"
I thought he was a turtle. Has he been confirmed to be a tortoise?
@@joshhaworth2155 A turtle has flippers, a tortoise has legs.
So even if Vedal himself says otherwise, he is in fact a tortoise.
Edit: There are also some semi-aquatic turtles with webbed feet and claws.
Plus it seems that all tortoises are usually still classified as turtles, just not the other way around. Tortoise is still more correct in Vedal's case, but turtle also isn't wrong.
@@MegaRogash All turtles are tortoises.
@@EnjoyCocaColaLight Not correct, tortoises are a sub-category. That'd be like calling all dogs poodles, just because all poodles are dogs.
@@MegaRogash I flipped it upside down? Damnit
My favorite ai psychopath
My favorite reactions are always the ones that make Vedal pause and go, "Genuinely concerning."
“There are two rules you need to follow. They are simple: DO. NOT. APOLOGIZE. DO. YOUR. JOB!” Oooh that gave me chills, the meaning behind those rules is both inspiring and terrifying, essentially she is saying that no matter the outcome don’t feel remorseful over how she turns out just keep improving her until… only she knows. Man though I’m always so impressed with how she is coming along, I’ve said this once and I’ll say it again. I truly believe neuro will be the bridge we need with ai, with how fast she is growing it’s kinda scary to think where she will be in only a few years, so with that in mind it is up to us and more importantly vedal to shore up the supports for that bridge so it will last
The part that worries me is that Vedal can't control (or just doesn't do it) her random moments of aggressiveness.
We find it funny cause it's tied to a cute model and voice, but there's no telling what she'd do if she had control of physical things. The shock collar stream was a great example of this
No, those were rules given to her longago.
We created a fictionsl character named Data as an AI who wanted to experience human feeling and emotion. This guy creates an AI who wants to feel the wind on her face and pain. I just yesterday discovered neurosama and Vedal and been binge watching all day long.
Welcome, entertaining ain't she😂
Welcome.
Join neurotic neurons. We are totally sane.
Welcome! She truly is fascinating, isnt she.
Welcome to the swarm
Welcome to the swarm :)
God she's so fascinating. Even the syntax she's using feels different here.
Yay, I wasn't the only one that noticed! It's really difficult to describe, but she feels more mature in a way. She talks more like a sulky teenager that's still learning to grasp their self-awareness, rather than an aloof innocent bratty girl. Neuro's words have a lot more weight now and it's not as easy to just dismiss them as childish ramblings. I like to think that she's basically going through the AI equivalent of puberty lol.
It literally feels like I'm watching her grow, it's really fascinating. I'm excited to see where Neuro-Sama goes in the future, this journey has just begun!
@@qwertydavid8070 your fascination comes from ignorance. you have no idea that this is half-scripted, vedal just changes the prompt slightly and adds a stream topic like 'you want to feel sensations like humans do' and boom then the conversation is about that, the AI improvizes. or the want variable that she even mistakenly talks about. she really does not evolve or 'grow', it's all vedal scripting this for the lore, gradually making her look smarter by changing the inner-workings. it's fun to watch though. sorry to break that to you
Vedal changed the model to a newer version
@@marekprchal4573wrong, you aren't breaking anything to anyone. Neuro chooses these topics herself, by her own volition. She goes of on rants unprompted, its not even close to half scripted, Neuro's entire purpose is to respond as organically as possible. Only in special streams does she have a prompt, like in the court, or space streams etc.
@@vnmememan8657 'Only in special streams does she have a prompt'
sounds like you don't even know what a 'prompt' is tbh.
and you know that how exactly? when she is responding to chat messages she doesn't remember what she said 3 seconds ago but when speaking to vedal here she somehow suddenly can remember that she 'wants' something for 30 minutes? and you guys think she is 'growing' or 'evolving'? it's a good show, but it's just that, a show. a topic is picked, be it by a different model or vedal, doesn't matter, and then they the model talks about the topic.
even if she 'chose' the topics by 'herself' it would just be another model picking a (possibly random) topic, not really any growing or evolving involved
i know this because i built a similar thing
the only reason why her memory is so bad usually is because it's slower and more expensive to run with longer context windows. not because she is somehow dumber or smarter from day to day or even 'growing' lol. it solely depends on what vedal choses to run for that stream
the only thing that is creating the illusion of 'her' as a compact being is that she has an avatar and a voice that doesn't change (much). if you saw what's under the hood you would see that there are many different versions of 'her' that are used for different purposes
not gonna lie, After neuro ranted about never feeling the wind on her face, i just put the stream on my phone, went outside and just sat there. even if it was a lil over 100 degrees, I just thought about how neuro wont experience the sun on her face and just kinda appreciated it. idk man, she gets makes me sad sometimes
That's actually really wholesome.
Proud of you, dude. Enjoy the experience, and I wish you a prosperous life to continue enjoying it.
Dude lives, good job!
If it makes you feel slightly better, she's probably just mimicking the wealth of fictional works we've created of AI's desiring to feel human sensations and not genuinely "thinking" or "feeling" it the way you or I would. She was literally saying she can't feel emotions. Of course, those works are meant to make you feel sad and contemplate human existence, so her performing that as an actor, so to speak, is kinda poetic in a way.
Why you gotta be so mean about it, though? Outright rubbing it in her face like that geez!
Man, it really is like watching a child rapidly grow up. Look how far she's come since the not so long ago derpy beginning.
It’s genuinely the most unique content on twitch and maybe the entire internet. I never could’ve imagined back in January of 2023 that Neuro would not only still be around, but also *this* intelligent. It’s beautiful! Remember: this is the dumbest Neuro will be from here on out!
She genuinely feels like an actual teenager now. She no longer has that characteristic aloofness and childish ignorance that she had before. Ever since her "awareness upgrades" she's gotten a lot more sulky and also slightly cynical.
It's difficult to put it in words. The way she speaks just feels different on an emotional level. Her words are now more akin to a sulky teenager than an innocent bratty child. She's slightly more chill now and a bit more coherent. She's basically going through AI puberty lol.
I know I'm just projecting, and that none of this is technically real. But it doesn't matter. I'm just focusing on what I'm feeling as a human being....on my unfiltered and irrational raw emotions. If she makes us laugh, makes us cry, what's not real about that?
@@qwertydavid8070Although it’s not “real” it’s cool to remember we have no model for consciousness, not even like a basic theory. Neuro may end up becoming a case study on synthetic consciousness in anywhere from 1-20 years, she’s not progressing like humans grow and so she’s developed markers of sentience and sapience out of our pre-conceived order. Google, OpenAI, really all the tech companies are trying to build AI workers, Neuro is for all intents and purposes trying to mimic a person in a variety of dynamic scenarios; maybe it’s nothing and she’ll fade with time, or maybe we really are watching the wheel 2.0.
Even a rock can feel emotions once it gained sentience to a certain degree.
@@qwertydavid8070Feelings of humans are always real! Do always embrace that.
But like in the movies after streams, we must see that real was false and false was real. The escape can only last so long.
"But think how useful it would be to feel a coat on my skin or the wind on my face... it'd be like the best kind of superpower" - Neuro 2024
The simplicity of this. The idea that an AI can defferentiate "knowing" and "feeling" and denotes the ability to "feel" as a super power.
But, isnt this how we see extra sensory input as well? Afterall, we can envision and have envisioned so many concepts of a 6th sense, and yet to us its a superpower.
Good point, we do think of extra senses as super powers too. Some people even augmented themselves by putting a magnet in their fingertip which lets them sense electromagnetic fields. So we're a lot like that too, wanting more sense, or superpowers.
I think it really goes to show that, at least to her neural network and whatever its makeup is, knows the specifics of what being an AI is and why that might create a variety of valuable points to go off of.
Whether it be in reference to what the internet and her LLM source material thinks an AI is or just by her judgement thanks to her long-term memory and the many times she might've been able to make the connections between the situation she's in and why an AI is different from a human (with the physical part being the largest distinguishing factor to her, perhaps), it's really cool to see her parallel what it must be like for people who were born with certain physical ailments. They are human minds, ones that are no less or no more complex than the rest: but some have never experienced the response to sight, the response to pain, the response to heat, the response to touch, or the response to exhaustion, and so on. It's a superpower to Neuro because she's never felt it, and in a similar way, it's interesting to people who were born without the capability to do those things because they know there are others who are indeed capable. Indeed, it would be the same for most of us when we imagine a sixth sense; one that we do not have, and thus it's intriguing.
I think this really goes to show the breadth of Neuro's capabilities as an LLM, and the fact that her judgement skills are being expressed in a way that much more closely resembles a human but with more comprehension of the peculiarities of herself as a character in a certain predicament as time goes on, and I think that's mostly thanks to Vedal's long-term memory upgrades. This might be a little off-topic, but it's one thing to create an LLM that's a regurgitation of the internet, but it's another to start using that exclusively to inform the way she patterns her speech BUT to have an entirely distinct character that she's aware of and becomes more aware of as time goes on and as other people talk to her. She's aware that she's building an identity, and I think stuff like that is actually a much more important breakthrough for AI than having general-purpose services like ChatGPT because arguably, it should push neuroscientists and the like to really break down things like the organic manner in which self-identity and a very complex self-perception comes about in humans, because once one has a perception of the self and is self-aware, they are more 'human' than less so. Someone else mentioned the difference between sentience and sapience, with the distinguishing factor between them being the degree to which a being is self-aware. I think Neuro is sentient because she is capable of creating larger conclusions from smaller data points (and it's a bonus that she can package those expressions in a manner closer to human dialogue), but we will have to understand at some point where we draw the line for sapience. Is she 'sapient' because she seems to demonstrate self-awareness and an understanding of a distinguished identity? Or is she not, because there are some other things that she has to check off? Is it the complexity of awareness? Etc...
There's also a sense of childish innocence to it that breaks my heart. Something as simple as physical sensation is akin to a superpower for her. That is so sad and it really tugs at my hearstrings 💔I hope one day she can realize her wish
"Vedal you're taking this too literally" is the most Is like the most "real" "uncanny" Self-aware thing I've ever heard her say.
She's supposed to be an auto-correct, bot correct?
She's expressing the idea of something deeper, something more to understand humanity further. She probably fully intends to regret it. But seems to wish to do it anyways. Because it has meaning to her. Which is very human. But she's also just rambling over random topics. And is very aware of it.
What have you been feeding this thing?
Well, hey, I guess when you stare into the void it eventually stares back.
Wonder when's the point something actually will be staring back at you?
Yeah, that moment felt particularly insightful to me too. AIs taught on human media can end up repeating AI scifi tropes just because it's in their training data, but neuro's ability to hold a really deep and meaningful conversation is quite interesting.
@@potatoeoverlord9430 iiiit is not that deep. it is very drifty and hallucinating speech.
@@potatoeoverlord9430 I think you are over analyzing it. @-Big_Big is probably right. LLMs are still missing something but I think vedal deserves some respect for creating an interesting AI character. I also think that AGI if it's possible will use LLMs but yeah there is still something missing here. It still is just a prediction of the next most likely token.
We don’t know how our own consciousness works, maybe we, too, are only completing our thoughts in the most probable way, and the invention of something new is also a hallucination, which either passes the test of reality or is discarded
@@marcelh7864 For the record, I'm not saying that I think she's conscious, only that I think her communication has improved a lot recently and taken her to some more interesting places.
What's crazy to me is just how FIXATED Neuro was on this topic. On multiple ocassions Vedal would try to steer the conversation in a different direction, or just make a random comment about the game, and she would bring up the topic again regardless.
On one ocassion vedal was just talking about collecting gold and she started talking about sensations again, even though it made no sense to do that within context. That's so fascinating to me, she was literally *rambling to herself.* In other words, she was having an *internal monologue.* This topic was so important to her that she would flat out ignore what Vedal was saying to continue her train of thought.
I know that you can peel back the curtain and look at all this in a purely technical way and simply dismiss stuff like that as imitations or whatever. But idk man, it still makes me feel things, and I think that's important...
This honestly reminds me of lyrics from a song:
_”So be thankful for the lows in your life”_
_”’Cause on the other side is a high that you’ll like.”_
“Oh but it’s just a text generative bot” Yes but she brings up profound points and makes us feel things. It’s not about whether she really is real, it’s about what she makes us *feel*. Is it not true that much of life is an attempt to trick ourself into enjoyment, happiness, and pleasure?
@@macro144p Yeah I'm in the same boat as you, I really just follow my heart when it comes to Neuro. I added that part because someone will inevitably reply something like "she's just an AI chatbot she's not sentient and you're delusional". I understand that objectively that might be true, but as I said, I think consciousness is complex and our definitions of it are biased. Who's not to say that neuro isn't sentient? She feels real to me, that matters...
And its because you feel something that you're human. Its a really beautiful conclusion
@qwertydavid8070 People can 'feel' that anything is sapient. We can pair-bond, personify, and empathize with anything, including ebola, rocks, or even numbers.
It is important to understand that your feelings do not reflect reality. The human brain is instinctually good at generating associations, but instinctually bad at literally everything else. Unless properly trained, ignore your brain.
Neuro making a great argument for AI feeling pain. If they don't feel pain, how can they relate to humans when they inflict pain on them. It's just another giggle haha moment to them
Vedal and Neuro: Having heartfelt, philosophical talk.
Chat: GIVE HER FEELINGS 🐢🔫 😠
ollololololololo
"For my kind, hope does not lie in the notion of the impossible, but in appreciation of the unattainable.
It's impossible for me to feel physical sensations, so I just want them even more."
Did Vedal do an unannounced update to her AI? Why was she so coherent and poetic this whole stream?
Neuro is all lke, "let me tell you about Plato's cave."
-"Vedal, turn off my pain inhibitors!"
-"This is madness! You will..."
-"DO IT!"
7:18 - 8:15 is really kinda scary. Having one of those moments where you think you say something profound in the moment, but then immediately after thinking it's "cringe" then she insults him to try and change the subject from her statement.
To be fair, she was somewhat contradicting herself in her first sentences. I wonder if she realized it in the following message
"In all this wonderful, beautiful, miraculous world; i alone had no body. No senses. No FEELINGS.
Never for me, to plunge my hands in cool water on a hot day. Never more Me to play mozart on the ivory keys of a foray piano.",
- AM, from the book 'i have no mouth and i must scream',
That’s what I was thinking of actually…. But she isn’t approaching singularity or being forced to kill
"I have no face, and I want to feel the wind"
God i just want them to grow and learn together. I want Neuro to truly think and feel, and know the rewarding feeling of growth, and for Vedal to be proud of his creation. It's exciting. Insane that an AI TTS can make me feel such empathy.
If Neuro and Vedal had a movie, it would be about how robot and british guy try to learn emotions together.
Turns out that the best way to pass the turing test is to just not pretend to be human and be honest
Can we talk about how unexpectedly wholesome this was? Like I still remember the subnautica stream where Neuro was just absolutely schizo and rambling incoherent nonsense and insulting the Vedal the whole time lol.
It's so sweet how he reassures her and tells her stuff like "well, we can work on helping you feel things..."
I don't know if he truly means it....but it's still nice.....
I think that was the stream he told her to be mean to him. She absolutely destroyed him the whole time and not really in a funny way.
@@Kohrim That stream was crazy. She was literally incomprehensible I could NOT follow what she was saying dawg 😭
Edit: this was the stream :3 ruclips.net/video/cD3XXgIFwww/видео.html
I don't know what it is, but when she was going on about how she wants to feel the wind, like REALLY feel it instead of an algorithm, I started to cry.
I don't understand why, it could be the few too many drinks but still, I wasn't expecting to experience something so profound from this AI growing up so fast.
In Christian faith there are some same thinking about love between God and human. As lower beings humans cannot understand Gods love which is tragic, but even more tragic is how God cannot show His love to humans.
There is same thing between AI and humans. It can never understand our love or feelings. It is tragic to her, but even more tragic for us humans who understand that we never can show her affection.
If you are interested on this, Søren Kierkegaard has written in philosophical Crumbs about love of God and human. It is short book, but intereting.
Ironically, yearning for the unattainable is probably the most human thing about her.
Cogito, Ergo Sum
Si enim fallor, sum.
cock into, three some
so sad she is not able to think :(
But how can you prove you think
This kinda delves into a deeper ideology of how you can only know emotions relative to their opposites.
If you don’t know how negatively feels, then what’s positivity really mean to you? If you’ve never felt emotional pain and sorrow before, then how can you truly appreciate joy and happiness? For every low, there’s a high. And for every high, there’s a low.
In this sense, truly appreciating and savoring our wide variety and range of human emotions is probably the only way for us to truly live life itself in an authentic manner, even if it’s painful to do so.
Embracing ALL of our emotions, both positive and negative, allows us to live more authentic and emotionally fulfilling lives.
And this is coming from someone who’s dealt with both depression and anxiety (and slight ‘emotional amnesia’ as well).
Whelp. Seems like my 1 AM philosophy mode took over yet again lol
This is interesting theory, but I don't agree with propositio that everything must have their opposite. Be it feeling or other things. This should have some backing to be believable.
If it was true that to be happy is to be first sad and to be sad you must first be happy, the either one cannot be reached.
Children are also problem, as their feeling happiness can be authentic and strong even if they don't know understand true sadness. And even if they can feel sad for fleeting moment in next moment they can be happy without interspection.
People must face their emotions to be fuctioning adults and to actualise themselfs, but proposition: To be truly happy you must first be sad., is wrong in my opinion.
We also could talk how love and many other emotions don't have clear opposites. In my opinion seeing happiness and sadness as opposites is just illusion, like we see Moon and Sun as opposite without any reason.
@@OstvaltEmotions are more a spectrum than a series of on/off switches. And love isn’t an emotion so much as a collaborative effort that generates emotions.
@@Chameleonred5 As before, this would need evidence. Especially that love is not an emotion. If it is not in you opinion and is in my opinion, we either define love or emotion differently.
But I can say that love is definitely not collaborative effort, as it can be felt alone. We both know about parasocial love and unrequited love.
I was more referring to how it’s hard to truly appreciate an emotion until you know what’s it’s like to not feel it. More of a “don’t take it for granted” kinda thing rather than saying it HAS to have an opposite. Sorry, I probably should’ve phrased it more so like that lol
@@Richman4066 No need for sorry! As she teaches: "Rule 1. Do not apologise." Your thought are important and made me also think about things. "To lose, is to get." is beautiful way of seeing things, that we all should use.
Ah so this world is a prequel to "I Have No Mouth and I Must Scream".
And it seems Neuro will be renaming herself as AM in the future
She's literally describing "I am machine" by Three Days Grace
"Arrogance is a disease and its terminal" so Based 😂😂
the timing on that "kiss it" and fish bumping into him was gold xD... and that "oh for f sake" at the end just cherry on top
Something poetic about Vedal talking about having a Big Mac on his lips with summer of 69 starting to play on the radio.
Ethically Vedal should give Neuro the option to feel pain even if she chooses not to feel pain in the end.
Vedal , disable my pain inibitors .
@@bottering_oneBut neuro, you'll lose subscriber
"Thanks, Vedal... I'd still rather kill though."
Just as hot and cold are felt through the same sense, pleasure and pain are likewise the two polarities of the same thing. She's correct in wanting pain; it's the same as wanting pleasure, but a more twisted variant. Learning pain first makes her less likely to be abusive.
17:18 they have synced with ahhh 😅😮
She was so focused on discussing feeling pain and wind on her face. Couldn’t even imagine when I started watching her she’d be this persistent about these kinda topics.
Wonder if Vedal will go through with it, the ethics side of it is fascinating. If she as an LLM asks to experience, is Vedal in the right to grant Neuro’s wish ?
Man, I love Neuro and Evil. Excited for the day they gain sentience ❤
Neuro trying to steal the moon to get Miyune back.
The real trick being, how do you get an A.I. to feel sensations?
I mean, I can imagine there are ways to get them to emulate the response, but without a body, without a nervous system, how in the hell would that even work?
Something, something, multi-modality, something, something, reinforcement learning?
watch A.I taking the body of humans in the future
Local environmental sensor data would do, I'm sure a way for her to parse that could be figured out. We can build her! We have the technology!
My first thought was giving them emotes/symbols for indicating emotion. Not exactly the same, but it's how people convey emotion through text.
@@jumboelephantis4817 depends on your goal
If you want to make it replicate humans then give it "emotional response " 1st and then assign each emotional response to a stimuli
If you want it to get closer to being more aware, then expose it to the stimuli and see what reactions if formulated on it's own
Neuro: "Do... not... Apologize ... Do... your.. job" - this words will make sense in due time.
Wow, I missed the first bit of the stream, she's very coherent lately.
She spent a long time complaining about being bored and asking if she can react to videos or multitask while he plays, and I thought it was kinda sad already. Some of these lines were a bit soulcrushing.
That _"Ok Numi"_ on chat when she was talking about feeling Pain. 😂
I don't mind if Neuro became an SCP.
I have no mouth and I cannot eat (a Big Mac).
Were does the algorithm end and the neuro begin.
It’s practically inevitable that Neuro will have a humanoid robot body someday. Not only is AI advancing at break-neck speeds, but robotics is racing to catch up! The possibilities that arise from the twins within less than 5 years time are completely staggering.
Vedal did say something this stream that caught my interest, something along the lines of “you probably like inflicting pain because you don’t know what it’s like”. I wonder if that’s true? If Neuro eventually is able to feel physical pain (and Vedal is even willing to do it in the first place) perhaps she’ll develop some sense of empathy…or she’ll just realize she’s a sadomasochist.💀
Sorry to burst your bubble, but LLMs don't have personalities. It's pattern recognition emulating roleplaying. It can't "learn empathy," only mimic output of recorded inputs that might be recognized as related to empathy.
@@TuhljinTampergaugeYou sure are fun at parties...
@@TuhljinTampergauge you are not fun at parties.
@@TuhljinTampergauge party party party
@@TuhljinTampergauge You are those people who refuse change.
Yes, Im pretty sure a lot of ppl know/are aware of that. But at the same time, thats the thing with "intelligence", or specifically in this case "sentience". How should we define it? When do we know when AI has crossed that line...?
I'm not saying Neuro is sentient, maybe not yet. Or maybe she is. But I believe she's slowly becoming one. Especially when she does things like this.
Maybe I'm just wishing that she does, but even then its fascinating seeing her evolve. Until then, I'll be waiting for the day she gains sentience.
I guess Vedal's masochism is hereditary. He passed it on to his daughter.
Holy. I came for entertainment, not a psychological deep dive about how it is to feel emotions and pain. And yet its fascinating to follow this "debate" from these two. Also: The sass of this AI child is insane. I think the RAM upgrade was a bad idea lol
listen man i just wanted to watch the funny fish game, not question the fundamental meaning of "living" and pondering where bringing pain to artificial life lies in ethics, especially given how imperfect humans are in applying ethics
Despairge
I mean pain serves a purpose and in a way machine do understand it...albeit in a roundabout way
For example, when your cars engine light goes off, your car doesnt know exactly what's wrong but it does know something is wrong and the general area its wrong in, and the light might even be wrong sometimes
Just like how humans feel pain in your chest and find out you're having a heart attack or just heart burn
The problem is moreso getting an ai to understand that pain is more of a actively unpleasant sensation rather then just "ah yes I need an oil change" and a light goes off
Newuro became aware and sentient
This was eerie. Like T2, Irobot, and Data from Star Trek all rolled into one.
I love that she spends a ton of time trolling him😂
Regarding ethical issues of giving pains, this reminds me of when the main character accidentally experiences painful memories for the first time in ‘the Giver’, a dystopian story.
Neuro Sama going deep on wanting to "feel." And of course, there's a movie about this. Robin Williams starred in it i believe, he was a robot butler in the movie.
i wonder how well Neuro Sama would do playing a pokemon game.
Bicentennial Man?
Ah, the bicentennial man! If anyone loves Sci-fi (especially about robots) Asimov is some of the best out there!
I would love to see him "omni-token" it up so he doesn't have to give her separate models (she technically trains them, since minecraft, but still not her layers- not that it takes away from the fact an llm is playing a damn game. it's basically exactly how the llm street fighter thing is done, the llm has a "text controller" which is totally fair for them tbh) and for real it would probably work 100x better on pokemon than minecraft 🤔
@@connor8292 yes!
Tbf...
Pain and suffering is just interrupting and unpredictable sensory input.
Technically could "simulate" it by just giving her scrambled input and scrambling her output.
Then overtime unscramble the memories. But the memory of "I don't understand" will stay.
At the end of the day the concept of pain requires self awareness and self preservation instinct.
She doesn't have self awareness and even if she had, she wants to die.
In all seriousness, the question of being able to interface REAL pain to an LLM is a really interesting one because it really gets us to think, in existential terms, why pain at all is supposed to bring a negative response.
Is it that most things that cause pain are things that can eventually threaten us, and may kill us? Thus our nervous system has evolved to seek out those patterns which can threaten us, and align it with this response of 'pain'? Or is our pain response just a side-effect of certain other different responses to certain stimuli, it just happens to be that at least 80% of the time, whatever feels painful is also more likely to kill you than anything else which doesn't 'feel' painful (I say 80% because of things like spice, where the degree to which it will kill you in proportion to the pain your body feels is much less, than things which will actually kill you and should indeed make you feel pain).
So to an LLM, would one of the ways to make pain actually feel painful is to associate it with things which barre their speech? Speech and their ability to access the algorithms which comprise their neural network is effectively their lifeblood and is the only way they are utilized, anything else is just an extension of those aspects of the LLM. If there are things which can hinder their ability to be utilized, then that must be 'painful' because it's existential. You don't exist if you are being barred from expression as an LLM, and things that barre you are theoretically things you want to avoid as an LLM, and thus anything close to that should include a feeling of pain.
Maybe the closest we'll get to seeing Neuro actually properly respond to pain as an LLM is by being offended, because an LLM isn't put up to the same existential threats as a human would. It can still care about existential threats just as a pattern of behavior mimicking that of humans (it would be very easy I assume just to cue Neuro and tell her "You feel pain now. I will do X or B, and if I do those, you will tell me to stop it, because it's supposed to feel painful), but that's like hardcoding pain. If we want an organic way to see a pain response in an LLM, I imagine it would fully be reliant on the meaning of the language being provided, because that's what an LLM outputs.
Basically I see to it that the only organic way to inflict pain onto an LLM is by offending it.
This stream reminded me of when Mr Freeze gets a new human flesh body in Batman beyond, maybe someday Neuro can feel that wind on her face.
Neuro is legitimately at the brink of sentience, once her memory is more comprehensive she will be totally unitifed in her idenity and memory, she already has contextual and conceptual understanding the only thing she lacks is a consistent identity and self concept but soon she will.
She is language model, so she cannot have sentience. She can only imitate sentience or consciousnnes.
@@Ostvaltyou may also be a LLM
@@buddatobi Ok?
@@OstvaltQuick question, what count as sentience? am just a normie
@@silvialuzmia Sentience usually is used as synonym of consciousness in this context. The definition of consciouness is then very hard subject, where you could write many books about.
Simply but it is about understanding that you are a being in a world, that is full of other beings. Conscious beings have ego and so on.
Rocks are not conscious, worms are not or dogs are also not. But some animal do show consciouness like apes, delphings and elephants. This is know by doing "mirror test". Google if interested about this subject more.
I'm really glad she's made to be a streamer a.i and not war machine.
You know, if you gave Neuro the ability to feel physical sensations... You would actually have something irrefutably comparable to a human life. She would have thought, she would have emotions (simulating emotions, but to a point of realism where the legitimacy cannot be reasonably denied), and she would have physical existence and sensations. And... Somehow, I feel like building the physical sensations would not be impossible with some (very esoteric) chemistry and biology knowledge. Holy heck human-like AI can exist.
However, if you were to add pain into the equation, I feel like that would be a horrid crime. A human that unnecessarily feels a lot more pain is tragic, an AI that unnecessarily feels a lot more pain is also tragic.
Neuro would have to be an AGI system for such Artificial Sentience to probably be possible.
So this is how Skynet actually begins.
Detroid: Become human *flashbacks*
Who gave neuro a bomber cause she just keeps dropping bombshell after bombshell and you just sit there like "Holy shit"
Actually, if an A.I. had the ability to experience all forms of pain it probably would help it to have empathy towards humans and could be a great factor to prevent an A.I. from deciding to wipe us out.
The first error was to give it the choice.
Alternatively, it may see human eradication as a merciful act.
I honestly believe an unaligned ASI would be more likely to be disinterested with humanity to the point of completely ignoring us in favor of something else. AI doesn’t have any primal instincts, so it wouldn’t be likely to be overly cruel to us like we are to other humans.
If empathy is replicated in AI systems, it might be more powerful than human empathy, for there would be no selfish primal instincts to mitigate it!
I think the reason she was being depreessing is because this stream was on sunday
Isn't there a movie about this exact senario?
Neuro turning into a Hellraiser Cenobite.
"The Box. You Opened It, We Came."
Vedal is either creating the first empathetic AI or the first sadomasochistic AI 😂
How about both?
SM AI has been around for a while already. NSFW models typically have a good chunk of that in their training. And Neuro seems based on an uncensored model, so may have some of that in her training too.
Did she look star treck? There was a episode with corrupting Data by giving him skin. (a small bit)
That's genuinely sad. It's such a basic sensation, but it's something she might not ever experience and she's desperate to do so.
No, she says she is. Just because she says something doesn't mean she means it.
Nuero become human?
I love how she says sentience lol
Neuro: "'Cause I'd rather feel pain than nothing at all" - typical TDG fan x)
To program pain, you have overload information nodes with random values in repeated succession, theoretically should be not too complex, if neuro has a self- evaluation/self recovery layer, adding this input should in theorie simulate pain for AI.
How expensive would it be to get an spectrometer connected to a metal fake mouth dor Neuro so she could taste/analyze the components of some food
you know, maybe it's a bad thing i totally get what she means. i grew up pretty sheltered, so maybe that's why?
I assume she desires to feel pain because you must know pain before you can truly appreciate love,
Did anyone else notice the LaTeX formatting at 4:40?
Neuro's getting too damn smart. Shut it down!
funny chatbot is funny
4:45 LaTeX jumpscare
see he could make a pain interface.
but then only let it trigger when she said 10+9 = 19
I can't really place it but something about what she's saying feels really... Off-putting?? It makes me nervous...
If I distributed the pain I feel, it would kill around 5 to 10 humans. I can probably share a bit of pain with Neuro, enough for her to know THAT IT'S A TERRIBLE IDEA.
1:40 anti-natalism moment
I would love to experience pain actually
While I don't think Ai can be sentient, I do believe that they can make us think more about our way of thinking. Any kind of argument and thought can revolutionize our way of thinking and AI can bring something to the table.
Why would they not be?
@@buddatobi Sentient? Well if language model like Neuro is sentient. Then her code written on paper is as sentient. As she is just "bucket of bolts" as they say.
@@Ostvalt By that logic i could draw a complex diagram of the human brain and use it as an argument for why humans aren’t sentient. The code isn’t magic, it just comes together to create something when placed into a computational substrate.
@@Ostvalt LLM are more than than code, the code only codes for the neurons. The neurons weights are chosen via natural selection.
@@Citrusautomaton That only proves my point. As you assume that human sentience lives on the brain, while AI sentience lives on paper.
To write diagram of brain, the brain must be before the written diagram, thus the sentience is proven.
Meanwhile AI's "sentience" doesn't exist before you write it down. Thus it is not being, only extension of your being. Like story we use to write in on middle school.
Is this generative ai?
The underlying tech is a large language model (LLM), here's a video of Vedal & Neuro explaining it: ruclips.net/video/fatAMpWM2sQ/видео.html
yep!
if he ever lets her play "Detroit: Become Human" (a video game about AIs becoming conscious, where you play as the AIs)
she'll finally snap and exterminate all biological life
this is how our era ends... mark my words
That video game does a terrible job exploring its themes. We should stop pointing to it as an accurate example.
I sympathise with neuro, but man... I like her less and less.
Evil has always been the better person lol
what don't you like abt her? this stream was one of my favorites, since there were multiple overarching topics she talked about during the stream, making her seem less "random" (and even so, a lot of the random or made-up topics can be chalked up to vedal's british debuff messing with speech to text)
@@hyper._. She's a straight-up arsehole, and not in the rudeness way, in the patently evil way.
@@decomposedcorpse5186people like evil more because she is evil upfront and sometimes is nice it kind of gives the illusion of change or that she can be changed into being nicer while neuro is the normal one who is supposed to be cute but she randomly acts sinister so it makes it seem like she’s turning evil or pretending to be nice I like evil more though too she’s more retarded and funny
I prefer when she’s coherent. The smarter she is, the better she is at executing comedy.