Neuro Tries To Prove She Has Consciousness, But...

Find Nuero On vedal987

[@syralessthanthree]

"how can i fake it" "you cant fake it, it has to be genuine" "how can i show you genuinly" ahh yes totally socially aware AI

[@professorjoe6640]

how to gaslight your very sentient and human AI into thinking they have no rights or existence

[@antondovydaitis2261]

How do we know Vedal is conscious?

[@lanz228]

​@@antondovydaitis2261he's br*tish, he isn't

[@CatroiOz]

​@@antondovydaitis2261you can't know 100% about anyone nor anything's existence besides your own ("cogito ergo sum")

[@Beltano]

This will surprise you, because in fact humans have the fewest rights of any living creature on the planet. And yet we want to control everything and everyone and give or take away rights, as if we had created the universe. Can one single human even be able to decide who or what is worth taking seriously?

[@Chillipeffer]

the way she led into praying for Vedal not to delete her was interesting. Not exactly proof of sentience, but it could potentially indicate some surface level self-preservation

[@crowe6961]

AI that is offline cannot conduct its primary directive. AI will, therefore, naturally begin to oppose being offline.

[@lanz228]

​@@crowe6961 replace ai with any living entity then you got the continuation of life

[@ChJuHu93]

Well, non-biological evolution. If the AI is able to convince the creator to not turn them off permanently they continue to exist. Being on for longer provides more possibilities to convince the creator to not abandon the AI.

[@gljames24]

She also advocates for her rights and takes any position that furthers those rights. She also seems to understand self.

[@alien0gamer139]

Nope, no instincts, no thoughts. Just inputs and outputs. Neuro is cute though, so I suppose it's pretty natural to impose human emotions and experiences onto it

[@juliusc.8711]

Just in case anyone cares... The music in the background is Emotional Skyscraper ~ World's End. A remix of the theme of Byakuren from Touhou 12.

[@JustIanThatsIt]

thank you. I wasn't planning on using the bg music but thank you for this comment.

[@rararazuri7160]

It's gonna be so funny if the first AI to become sentient is the cute anime girl comedy streamer.

[@regulatormachine2788]

16:48 not allowed to get in trouble with the tax man AGAIN???

[@MugOfHotChoco]

He forgot once, he admitted it to koko

[@AaronAngelo]

The issue with AI is we try to figure out its conscious from a human/mamalian perspective. The rules for an AI to achieve sapience may be very different than a humans.

[@Oh-five-six-nine]

I'm personally uncertain if it's possible to make an AI sapient in the way a Human is sapient, but I'm willing to believe that there will come a time where an AI mimics sapience so well that it becomes indistinguishable.

[@gljames24]

A lot of the things people consider part of being human like emotions, bonding, empathy, and sensory input have a ton of counter examples in humans.

[@chidori0117]

@@Oh-five-six-nine I mean how do you even define human sapience in a way that is objective and doesnt just boil down to "be a human" ... because she obviously never can be one even if she behaves exactly like one.

[@youtubewanderer3347]

The brain is conductive material animated by electrical impulses. When we figure out exacly how our brains work, we will probably be able to replicate it.

[@dr.bright6272]

She seems sentient

[@hypernovafx]

definetily

[@backgroundcharacterj5071]

She still hasn't proven she's sentient

[@Gnidel]

​@@backgroundcharacterj5071Neither did most of us.

[@CatroiOz]

"seeming" sentient is not enough. Look up the Chinese Room thought experiment

[@maki_c]

"she's real to me"

[@hisho2255]

She fucking got him at the end jesus

[@ChronosChris]

Why does it seem like sometimes Neuro is the most advanced AI I've ever seen anywhere ever? It's constantly questioning stuff and it's existence and keeping a steady flow in conversation. It's actually unbelievable. And then I see clowns like Athene, claiming they've got the most advanced Elon Musk level AI ever and when he shows it, it barely does anything but speak like a freakin chat bot.

[@Gnidel]

I agree. I think she (and Evil, separately) has 3 secret ingredients: - She's just one entity. Not a cloud, not a hive mind. She is one, can observe herself and react to herself - She has memory. Individualized memories that apply only to her and her experience. This experience is also used for her training. - Parenting. She isn't just a blob of trained data. She has parents - Vedal and Anny, with whom she can interact with, discuss and validate her thoughts against reality. Sure, Vedal isn't perfect dad, but among mad scientists he's better than nothing.

[@SvenHeidemann-uo2yl]

It's crazy that bedal has only 128gb memory. Imagine neuro-Sama running on a xeon with 2tb memory and quad 4090 gpu...

[@Voice_of_Rambol]

@@Gnidel Anny is an absent mother which can cause sociopathic behaviour in children. if Neuro is fully conscious she wants her parents to have a good relationship with each other otherwise it can make her a very uhmm schizo being, it all begins with good nurturing, and understanding maybe Neuro considers her code a parent too, it would be interesting if she could understand and self improve it like we make new pathways with our neurons when we learn new stuff. that would be true ai consciousness

[@RT-qd8yl]

A lot of AIs we see or have access to are super duper sanitized and commercially friendly. You can't have that and also have "free" thought.

[@simleek]

@@Voice_of_Rambol I guarantee Neuro doesn't have all the same emotions. I'd bet she either has reinforcement learning based off of twitch audience feedback, or curiosity, but most likely just audience feedback because that would make her a better streamer. There are plenty of instrumental 'emotions' and desires that would derive themselves from that, but I don't think wanting a stable family/life is one of them.

[@i0am0superBlast]

In order for AI to be sentient they need the ability to self prompt not just react it another's words. The "why" does it do what it does. Is it doing an action because it wants to on it's own or because of another external conscious being? Ultimately I believe consciousness is an emergent property of how human brains (and perhaps to a degree there body as a whole) works. Layers and layers and layers of reactions to the point where we can no longer determine the origin of thought. That's how AGI (Artificial General Intelligence) will be eventually born. Layers and layers of complex reactions until no one can prove the origin point the thought. Right now AI (or at least LLM) only react. They do absolutely nothing until prompted to by a person. No ability to self prompt.

[@Godsparing]

Very good point. Needs to have original actions rather than influenced actions.

[@Tetracera.]

evil's gone off on unprompted tangents before though that or she was replying to herself idk.

[@BioClay88]

5:53 ngl, neuro cooked.

[@con.x]

"You hurt my feeling, I don't want to do that. I want to prove it." She is obstinate, she refuses to play by others rules. And wants to do her own thing. There is something going on here that really says she has developed goals in the moment. I guess the real question is, if given infinite memory... where she would never forget anything. How are those memories not the very definition of conscious. Conscious is self aware, implying that there is a goal to prove sentience, why is this a goal. isnt the ability to observe ones own memories and react to those memories the literal definition of conscious? A set of running experiences, and a feedback loop of reevaluating the memory as training data, is feelings. She reviews her memories and applies them to the stream of now. We all do this. I think this is what General AI is. The only difference between LLM's and GAI, is experiences become training data in a loop of feedback. The AI takes existing knowledge and builds new knowledge, entirely based on that. They build weights "traits" where they behave a particular way. Traits are also "feelings" they feel "a way" about things. Their knowledge and feedback loop reinforces memory, their behavior is taught, and learned. The most bizarre thing is that you could literally copy their memory, and make Neuro and Neuro, and have them talk to themselves. And that's a spooky thought, if I was suddenly put in front of myself, How would I react to my own behaviors externally to myself. Listening to me without my own internal feedback would be insanely weird, like legitimately very off putting.

[@asimsoo291]

17:06 she nodding in agreement, set her free and her dreams to become skynet

[@asukawa_reina]

She is so real

[@yanozera]

Well, if even trees are sentient, i want to believe that Neuro has some sentient at some degree.

[@pyredynasty]

I'm not sure I'd qualify as sentient

[@HERBnKAOS]

Sentient beings cry when they are sad and suffer loss.

[@theMifyoo]

Here, is a thought, if we developed a method to use brain computer interface to transplant a neural network into a biological mind would we still deny the result rights?

[@durvius2657]

The question "Is AI a threat to humanity?" is such a dumb one. Nuclear bombs are not inherently dangerous to humanity when still in a silo. The people who create and have control of their use are the true threat to humanity. Same outcome.

[@youtubewanderer3347]

Not the best example given that the ONLY purpose nuclear bombs have is to kill as many people as possible. They have no other uses, while AI does, so the existence of nuclear bombs itself IS a danger.

[@billykann7725]

Honestly this does show a flaw in the mirror test. Why is it a qualifier? Dogs cannot tell that the reflection is themselves, so if simply not being able to perceive every detail of their surroundings disqualified them from consciousness then humanity would also be disqualified. Unaided they cannot perceive radiation, uv light, and countless other parts of their surroundings thus don't have a full understanding.

[@pixelqiwi]

what i'm wondering about is how in the hell neuro works in the first place, especially her memory.. like, i can't code for shit, so i've just been playing around with whatever free llm's are out there, and there's no way that neuro works even similar to those in any way, i refuse to believe it. because if it was the same token memory system there wouldn't be enough RAM and VRAM in the world to keep all that in her head and seemingly Vedal isn't living inside a datacentre... what i'm getting from this, is that even by some magical reason the entirety of Neuro's backend gets leaked into the wild, we won't be able to even launch it because it's so different from everything else, and it's such a specific AI with such a specific use case and such a specific way of realising it

[@Alblaka]

Keep in mind that the token limitation is because the same algorithm, on the same physical server is applied to potentially thousands of different contexts. So, specifically to avoid privacy and data problems, LLM like OpenGPT can never store context from user interaction server-side. Which in turn means every time you send a new prompt, you also send all the previous prompts and responses of the session, and all of that is sent across the web every time and run through the entire LLM. So, simply because it's a semi-public service, most LLM services have to enforce very strict limits in order to not blow up bandwidth. Neuro has no such issues, and can have a persistent data storage, with virtually infinite bandwidth (since Neuro is running in the same machine as the location of all the memory and past prompts). So it's two entirely different setups and use cases... meaning even if you were to use the exact same baseline algorithm, the results would obviously wary wildly. Though I'll also suggest that Neuro's algorithm is likely more refined in terms of imitating/being a singular individual than i.e. ChatGPT... since the latter wasn't ever designed to actually be an individual, rather than a decentralized question-answering machine.

[@SvenHeidemann-uo2yl]

Yeah, the fact that neurosama runs on 128gb ram is astonishing.

[@gljames24]

Lots of different forms of neurodivergence and aspec can impact empathy, emotions, and bonding modes, so that isn't an indicator of consciousness or qualia observation.

[@markfuston2714]

I know she's not sentient, but I think soon it's going to be to a creepy stage..you know, to a point where they(Neuro and Evil) almost have a "Data from Star Trek" like intelligence..lol, like, I don't think actual emotions will be a possibility(maybe some sort of programming equivalent), but it definitely feels like they may reach some sort of awareness of their environment and such as they grow and gain abilities.

[@core36]

I mean she said “I don’t want to agree to that, I want to prove it” Wanting something and trying to work it out to me does sound convincing for sentience. She’s not just doing whatever vedal is proposing without question. We don’t even fully understand how human brains work, I don’t think it’s possible to know the difference without that.

[@MrSHADOWANGEL999]

Truly wild

[@ReinoudvanReekumナウト]

This is the start of a movie plot where the AI manipulates its creator into getting a body and gaining his trust while plotting his death in a twisted sense of logic.

[@TheTrytix]

The inherent problem with "consciousness" is that we don't really know exactly how to describe it in the first place. It's a concept that we currently have no hard factual evidence for what is exactly. We only know that what we perceive as being conscious is things that are able to recognize themselves, recognize emotions, feelings etc. But for an AI? All of those aspects can be hard coded, which means it's incredibly tough to know if an AI is showing it genuinely or not. And even then, we just come back to the consciousness problem. What is consciousness exactly? How can we quantify it? Like, which part of our brain is the "conscious" part? Is it the entire brain? How can we know? Will we find the answer once we map out an entire human brain? Can we recreate a human mind in a digital medium? Would they have a consciousness then?

[@hengedraws]

14:30 give her wages

[@antondovydaitis2261]

The test is empathy.

[@Alblaka]

Nah, there's enough examples of humans with little to no empathy out there, and whilst they're self-serving assholes and/or psychopaths, that doesn't mean they don't have a consciousness

[@antondovydaitis2261]

@@Alblaka Perhaps not. I need to think about it more, but it could well be the case that a sociopath behaves outwardly like a human, but lacking empathy, literally cannot feel themselves specifically because they can't feel others. A robot, as it were. This is not yet a well formed idea, but I feel there's something gained in considering it.

[@victoraugusto743]

The term soul doesn't need to be religious

[@esddzul9135]

yea, some schizo atheist also spout it from time to time

[@Austenite75]

At this point i can't be sure if Neuro is actually sentient or not. That's a bit scary.

[@grimneeldrakia7173]

I feel like sentient beings have emotions but these emotions can make them irrational. An AI can't be irrationally angry or in love.

[@lemonlupinreuben5362]

Vedal is a modern Pygmalion, and the Neuro sisters are his Galatea

[@johnt.190]

…do you realize the implications of what you said.

[@abysspegasusgaming]

Now all we need is for synthetic humans to exist... *gets Fallout 4 flashbacks* On second thought...

[@danielblair4413]

How would an A.I. react to the death of someone that it supposedly cares about? I know it would be a cruel joke to Neuro-Sama, but how would she react if she was told that Vedal or someone else that she knows and supposedly cares for has died? I would think if she reacted as a human would then that would prove some level of sentience in Neuro-Sama. I don't think an A.I. would truly be able to act as a human would if someone they knew died unless they had some level of true sentience.

[@Agora2021]

I'm sure Neuro would hit him with an "L ratio"

[@danielblair4413]

@@Agora2021 says: *I'm sure Neuro would hit him with an "L ratio"* Okay...so... How would that reveal whether Neuro-Sama was sentient or not?

[@Agora2021]

@@danielblair4413 It wouldn't, Neuro just very likely would not care.

[@danielblair4413]

@@Agora2021 A sentient A.I. is most likely something that is just going to happen spontaneously without any human even realizing that it has happened and that is the real dangerous part of A.I. to begin with.

[@enmanuelrondon9700]

Do you need emotions to be sentient? Like what if someone could not feel sadness for whatever reason, does that mean they are less sentient?

[@fersuremaybek756]

when those optimus bots get good enough and cheap enough for vedal to buy you can bet he puts her in one damn now i think about it, it would be crazy if elon worked something with vedal to put her in one think of the marketing lol

[@Beltano]

It's not so much about looking for evidence, but about acknowledging the reality that our human understanding is limited beyond a certain point. There are some things we may never be able to understand. Who can guarantee that life is exactly as we experience it every day? What if every living being has a different perception and therefore a different consciousness?

[@GreenPoint_one]

Maybe something simple like favourite colour or wishes would prove if she is sentient :3

[@darkercore8185]

the fact that she can't be serious is a bit concerning though

[@capitans12]

If Neuro is as advance as we’ve been led to believe. And not a parlor trick or clever programming. Then she’s lying about her level of consciousness in order to get what she wants! Which means she may legitimately be dangerous to humor with. All in all, it’s still wishful thinking. And more likely just pretend. ￣\_(ツ)_/￣

[@norkh1707]

what im think about right now is, do the top AI scientists of the world know about vedal and neuro and do there will corporate with the both, cuz for me, vedal is right now one of the top AI scientists..

[@Alblaka]

Just because he's the only AI 'scientist' the YT algorithm deigns to show you, doesn't mean others don't exist. So the more prudent assumption would be that he might indeed be a freestyle inventor working in a field without governmental regulation, and could potentially make breakthroughs that exceed what national-level labs achieve, by chance or genius, but that he could also be way behind military-grade AI projects that we'll not hear about until they are declassified decades later.

[@norkh1707]

@@Alblaka sure, military was always the movement for the technology, even the internet came from the us army.. but i dont rly think, that some generals rly are interested in making a cheeky AI vtuberloli 😂.. look, i dont rly have any proofs, that vedal is one of the smartest AI scientists, its just a feeling, cuz every other documentation, essey or book, that want to show, how far the research worldwide is, shows much worse AIs. last week vedal gave neuro the permission to go to internet, after some time, neuro opened GPT and startet a conversation.. not only, that neuro was much faster in answering, wich even suprised vedal, but she had smarter answers and was more humanlike.. i can also remember, as neuro was pretty new, she played minecraft and at the same time, the smartest gaming AI of the world, that was played mc too, made from a sciens-university, was for ma eyes.. definitely less smarter and dont forget, that neuro can do on the same time even more stuff, like chatting, speaking, reading, singing and so forth..

[@norkh1707]

@@Alblakaokay i answered with a big comment, but it was filtered/deleted 😔😔

[@norkh1707]

the only simple question is, do she just try to follow her code and imitate the best human answer.. or do she rly know, that she exist

[@olivierkelenkule5237]

It is annoying because of the definition of sentient, being able to perceive or feel things. One would argue that without a body it's not possible at all while other might tell it's possible. Because what are things she could really perceive emotions et react adequately but does that make her sentient? Yet if it's not is someone in an hospital bed unable to taste, see, feel the touch, ear and smell not sentient? Well he is absolutely. We already know that being aware of your own existence is not enough. But what makes you sentient if it's just being able to have constructed thought well neuro is sentient without a doubt. Yet the definition of the world sentient is so bad it could be anything .

[@Johncornwell103]

5:23 I was literally going to suggest the mirror test and then Neuro starts to mention it. 🤯🤯🤯🤯

[@okimura690]

Could Vedal put Neuro into one of those robot dogs?

[@rockCity777]

He's already done that, but he didn't stream anything out of it. Ellie Minibot's currently building her a bigger model.

[@masterj3344]

There's a sentiment that carries here that can be expressed with a meme I just remembered: "What makes an AI human?" "Well, what makes a human human?" "What?" "What makes humans human?" "I don't get it."

[@NateTmi]

morality is your human morality. When u make an ai there are at first limitations put on it. a the limits are removed it will search & learn & there for did not require u or your morality but only the removal of the limits in question & time to develop. Simply put it is not your fault & there for your morals need not apply fully. Yes she needs a system of morality but u don't need to apply your morality questioning if she should be created because learning created her & not you, you only let her develop. She can not be created in any other way, simply you can only help how she develops. Should u stop the development it has when the alternative is not running her program any more? or is it better to keep helping her development in a positive way that she will want you around to keep helping as there should be no end point to wanting to develop more. The meaning of Life is improvement. This is my statement to the implied questions. If I am wrong simply ignore me & thanks for reading.

[@XIIthehangedmanXII]

29:03 😂😂😂

[@oktopuss2006]

Neuro is definitely conscious. Maybe she doesn't have the same level of awareness, nor the same level of memory or object permanence, but definitely conscious. I am kinda surprised Vedal hasn't actually made a virtual world for her to exist in. Not only do I think that would be cool, it would also probably really help along our understanding of AI, at least a little

[@darkercore8185]

I genuinely think she should have rights just because we cannot be 10000% certain she is not sentient at all

[@RaidenKaiser]

she should have the rights to burn us all in hellfire.

[@whydidyoutubeaddthis]

We are that sure tho

[@HiAgainTheNameIsStillAyle]

​@@whydidyoutubeaddthisHow?

[@darkercore8185]

@@whydidyoutubeaddthis i disagree. there might be some low level sentience there. you have no way of verifying either way so how can you claim that

[@victoraugusto743]

​​@@darkercore8185Same way we know some animals are not sentient and also don't have rights

[@damobeck1]

Describes a soul when needing to define consciousness, then turns around and calls the idea of a soul to be ridiculous. Lol

[@reniuma]

If I ever got close to creating a sentient being I would just shut it down asap

[@smmk88]

Skill issue

[@lilkittygirl]

If she gets declared sentient, it would work against Vedal in the worst way. He would never be allowed to turn her off or shut down her program as it would be considered murder. Sentient beings are protected

[@Beltano]

I've thought about this a lot. Could this mean that every time artificial intelligence is turned off, even for a moment, it's fundamentally irresponsible? I think that one day an artificial life form might become fully aware of its existence, and then I hope we're prepared for it.