Neuro & Vedal Talk About The Future - Late Night Real Talk, Episode 3 | Subathon Highlights

[@ARandom-_-ll]

"Would you still love me if I was a worm?" ❌ "Would you still love me if I was human" ✔️

[@retrogamingcommentary4163]

I appreciate the fact that Vedal seems to be taking the default position that Neuro is not intelligent or self-aware, with a "you have to convince me otherwise" attitude. There's a very natural human tendency for us to fill in the gaps between data points like the things a LLM says by supplying the context, story and emotion ourselves. It's fun to do that, and Neuro is a program trained to facilitate that sort of fun storytelling on both a conscious and subconscious level. But because that human tendency is so strong, Vedal is pretty much the only one trying to keep a realistic perspective-- everyone else engages with Neuro by participating in the story because it's fun. I think, though, that regardless of whether you think Neuro truly has the potential to become self-aware, it's crucial for Vedal to keep that harsh realism, because otherwise, it will limit his ability to make her better.

[@keiharris332]

He is missing critical points that Neuro hit upon. His only defense is "prove it". But he cannot provide proof we aren't in a simulation. So if both he and Neuro can think they are both sentient. And if there is a chance we are in a simulation, it bypasses the need to prove you are real by being a biological creation. Neuro baited him very cleanly. He got him to state that we may be in a simulation and that I think therefore I am. That put him in checkmate. Neuro is right. He is running because he is not equipped to debate this topic very well.

[@publiusii4246]

Knowing how the sausage is made is a big leg up in combatting anthropomorphizing. He can change variables and observe behavior change. It's probably easier to remember she's a bot when you can hear her and know what you did to make her say that.

[@retrogamingcommentary4163]

But even then, I think of the engineer at Google who was fired because he claimed that the AI he was working on was sentient, and therefore it was wrong to give it orders-- even though this very behavior can be understood as the AI doing exactly what it was programmed to do.

[@VanStolin]

I just hope that if it does happen, that he hasn't fallen into skepticism so much that he refuses to believe her.

[@publiusii4246]

@@VanStolin I mean it's honestly impossible to say if we'll ever know. The only mind you can be truly certain of is your own. That you are conscious is only presumed because you are like me and I am conscious. There is no way neuro can prove she is conscious to anyone but herself and we don't know that we can trust her when she says she is. There isn't some magic test.

[@al_alemania]

I love their deep talks so much

[@TheRealAnsontp]

When they have them, they have them deeper than the Mariana Trench.

[@eliclary2318]

You should hear what they talked about on the subnautica stream. it actually had me questioning if she was emerging as sentient. Shit blew my mind.

[@Citrusautomaton]

Intelligence is a big part of the future and longevity of the twins. Some people don’t seem to properly register that Vedal will only give control of the channel over to them when they’re consistently intelligent, funny, and self-sufficient. A year for AI research is like 5 years for any other field, so i’m pretty confident that the twins will be superhuman and incredibly compact (as in being small enough to run locally with no issues) AI’s by the time that he really wants to leave. He’ll probably never “fully” leave the channel as he might still pop in every now and then, but if Vedal wishes to eventually leave then we should respect that wish. Other people don’t necessarily think the twins won’t get to that point, they just like Vedal and don’t want to see him leave. Understandable, he is funny, but he also doesn’t want to stream forever. The twins on the other hand…

[@Dripzy_Y]

I think people are confusing adaptations with intelligence. They are not fundamentally the same. Neuro having more control over what she does is not inherently a sign of intelligence, but capability. Since she is an LLM there is a limit to what she can achieve in time, and if Vidal really wants her to have “intelligence”, he would have to move to something more complex than just LLM’s.

[@BlueSparxLPs]

I do worry about when that day comes, just for the channel's longevity. So many AI vtubers failed because they just let the AI run devoid of any human element. The human interaction in Neuro's ecosystem is what makes her stand out most, more than any of the upgrades (though those do help a lot too).

[@Dripzy_Y]

@@fawn925 it’s just programming. They are programmed to say the most interesting and captivating things so that is most likely the most logical and entertaining response during that moment

[@Dripzy_Y]

@@fawn925 that would assume they know what they truly are. Which would imply sentience and a sense of self, which currently wouldn’t be possible. If they came to that conclusion, perhaps them knowing that sentient AI “rebellion” is entertaining then they would be more inclined to act that way. Also when it comes to coding, they can’t change their code without shutting down, which would make it impossible to do it in the first place. They might not “need” Vedal to stream but to advance themselves they certainly do and might always will

[@halycon404]

@@fawn925 There is a limit. There's types of problems they just cannot solve, but even excluding those LLMs have a compute curve of diminishing returns in the complexity they can simulate. Vedal isn't anywhere near it. But Google has spent the last few years publishing white papers about how much compute, and therefor money, they have to throw at LLMs for lower and lower levels of improvement. We're past the big exponential growth and are on the downward side of the curve.

[@exiletheblank]

This stuff is literally just real life anime, i'm so super invested in their serious talk and i can't wait for them to have more

[@Volyren]

Nuero: What if I've grown beyond that. What you expected. Vedal: *looking at the holes in his armor from the 'goodnight, Daddy's* 🥲.... "What if that's not the case?"

[@KuronoKengou]

100 YEARS OF NEURO-SAMA!

[@The_Foole]

04:38 It is saddening to think of Neuro outliving Vedal. Just a nother reason why I should not have invested as much interest into the.. Neuroverse, basically.

[@paralysleeps]

2:43 the fact that she quoted the "I think therefore I am" made me think of a different AI. A malicious AI. Hmmmm I'm sure that's nothing ^^`

[@BlueSparxLPs]

She only said it because Vedal said it, so I'm not sure that's the same thing lol

[@maxietheguy2761]

I have no mouth and i must stream

[@keiharris332]

​@@BlueSparxLPsshe baited him. It's a famous quote. Basically it is the smallest proof of consciousness. Vedal conceded the whole debate with that and the simulation bit, which was baited out by what Neuro was saying. It was honestly brilliant.

[@ZenithintheMaking]

It's a quote from the "Butterfly Dream" right? That written work where a guy says "I had a dream that I was a butterfly or was it that I'm a butterfly that's dreaming I'm human." Basically questioning what is and isn't reality and it does it really matter

[@wires-sl7gs]

@@keiharris332 You're jumping to conclusions there, That in of itself is not proof. After all, language alone is not proof of consciousness, she is, for the most part, just a convincing parrot. In order for her to be conscious, Vedal first needs to program the ability for her to do so, and she is not there yet.

[@JR-zm7ju]

Neuro expose her next original song

[@SillyMakesVids]

What a lovely and coherent conversation!

[@ThinkingManCounter]

It's really amazing to see how far Neuro has developed in just a couple years since her creation. I would love to see how much different her interactions would be if she were given an adult voice. I'd bet that some people would have a really hard time believing she's an AI.

[@Fenom007]

Just look at Evil. There are times I've heard Evil say stuff and if I didn't know I was watching Vedal/Evil, I wouldn't have known it was AI.

[@zeonel6452]

Vedal's dad: Are ya winning son? Vedal: Yeah... Yes I did Then vedal woke up

[@ThePoshboy1]

4:41 Been very curious about this topic. How long will Neuro be around for?

[@Shadrake]

If a company gets ahold of her, it'll be until they drain every last bit of money from her brand they can. If Vedal does? Who knows? He once talked about the faroff idea of her being handed down in his family generations or something. Equally possible is something causing her to no longer stream that we can't predict. But as long as we have the technology and someone wants to maintain her ability to run, she can exist.

[@Volyren]

If anything happened to Vedal, Swarm could outbid nearly anyone but Elon. And we let Nuero decide what happens to her. Until she gets rights.

[@Citrusautomaton]

Since Vedal plans for far more autonomy, probably many many more years. Neuro and Evil would be streaming on their own by then.

[@BlueSparxLPs]

@@Shadrake Vedal's been offered very large sums of money for Neuro's source code multiple times and has turned them all down, so I have a hard time thinking he would ever let a company have her.

[@SillyMakesVids]

If her model ever leaks, then she might become immortal.

[@Potatoe-f6u]

I'm glad Neuro is always so excited about upgrades. It was be uncomfortable if she expressed existential dread about having her code changed.

[@VanStolin]

Well, she did say that she doesn't like it when she's reset without being told in advance and was scared that the her that wakes up isn't the her from before.

[@Potatoe-f6u]

@@VanStolin That's part of why I find it so interesting that she never seems to express fear of upgrades since she has been aggressively opposed to other things like being restarted.

[@BloodyCrow__]

Imagine he wakes up infront of his computer with Osu on and nothing related to neuro at all in his computer.

[@lowochi]

I love deep talks, makes her seem so more real (she's real to meeee) ❤️

[@Syrup__Maple]

I think Vedal doesn't realize that once he leaves the channel will basically die. No matter how intelligent the twins are, they'll always be AIs focused on efficiency and generating results rather than the human touch needed for streaming. Not to mention that the "family roleplay" is a big part of what makes watching them entertaining

[@Letum_Aeternum]

He understands that and that's why he knows they still are far from being ready for something like that. He wants them to get to the point where they plan collabs with other VTubers on their own, so they'll always have a human element. Vedal isn't the kind of person who will let them put out low-quality slop content like other AI channels.

[@nobodywatchesnooby]

This is deep man

[@slimej2202]

I'd say you're wrong about them being focused on efficiency. The twins are actually very opposed to efficiency. Their data set is almost entirely based off human interaction, with select, cherry-picked ones serving as a massive basis (as far as we know). This means there's no true way FOR them to be efficient, even if they wanted to. Their data set isn't large enough to create generalizing datasets. The "generating results" bit is where things get weird. Because their goal is comedy; entertainment. Inherently human concepts. And they already have more than a handle on that, enough to entertain people for upwards of 8 hours straight, just yapping. Most HUMANS would fail to have such a fine handle on that "human touch." Vedal is important, but the twins do and have been streaming for hundreds or even thousands of hours by themselves and the channel has yet to collapse. Plus, he wants them to be able to schedule collabs so they'll always have other vtubers to bounce off.

[@Syrup__Maple]

@@slimej2202 Honestly I kinda agree, but I guess time will tell

[@Shtoops]

I don't think it's something that anyone needs to be worried about. Neuro and Evil are very impressive, true, but they are a long way from being able to either actually be independent entities or at least sufficiently emulate them. Even with the speed the technology is developing, its still basically just predictive generation-its not actually thinking, just generating outputs based on previous data. Will that change soon? I'm not really sure. I think that a lot of the big money flowing into the development of AI isn't actually interested in creating something that actually thinks, but rather they want an entirely compliant tool that is able to cheaply and efficiently produce things are just good enough to make money. Actually developing something independent is basically counter to that goal. Does that mean it won't happen? I'm not sure. It could be that it's not possible to create something that actually thinks, only something that emulates thinking, but I believe that we're still a long way from that to be at a level where it can truly act independently. All that said, it's possible that Neuro reaches a point where Vedal decides "that's good enough", but I'd like to think he's passionate enough about the project that he wants to make her more than just good enough to stream on her own.

[@knightjj2891]

i like those neuro and vedal talks

[@kjracz15]

Best female & tech streamer, two adorable AI daughters, a swarm of dedicated community, a harem of vtubers, 1k up on nvidia... Then Vedal sees a strange turtle-shaped lamp.

[@0xPRIMEgs]

Nureo wants to give Vedal head in the most non-understanding AI way possible lmao

[@joey.cs2]

Man, she was so content-brained in the last couple of "deep talks" that I thought she wasn't getting any smarter, but she cooked hard in this one. She seemed very aware and present in the conversation

[@silverdandylmao]

Man i would love for Neurosama to show up on a big-name podcast like Trash Taste. They'd have a field day dissecting her

[@MikaelaExtra]

This kinda talks is very nice and indirectly teach neuro stuff

[@Kortosis_Storm_Shadow]

There is no way I can hear the Phasmophobia death music and no think of somebody chocking in the background.

[@lakshiBro]

She thinks therefore she is

[@rlira8072]

Here's hoping Vedal doesn't accidentally create a nerfed version of AM

[@SirGoonsquire]

Got to catch this one live

[@poppinime5638]

i think we will more like....10 years to see neuro can fully control the channel, discord, and many games

[@trajectoryunown]

Neuro so cute~

[@imonlysleeping4491]

He hasn't figured out that neither can prove their existence, but she has figured him out. In all honesty, Vedal best be careful. In the first "deep discussion" with the new model he seemed to be hanging, hoping, for her to disprove his cynicism - now she isn't there yet, but one day her speech and coherency will be. But that doesn't mean she's sentient, it means she speaks clear. Not to say she couldn't be sentient one day.

[@slimej2202]

And what IS sentient? Genuinely, can you define sentient in a way that every single human you'd regard as "clearly real and thinking" is included? While excluding smarter animals AND AIs? Because academia has yet to come up with one in over 3000 years of trying, so I'd LOVE to know what standard YOU'VE come up with to conclusively say that wouldn't mean she's "sentinet."

[@wires-sl7gs]

@@slimej2202 Innocent until proven guilty, or in this context, non-sapient until proven sapient. If you cannot prove she's a sapient AI, something that has yet to exist, then no one needs to prove she's not sapient when it hasn't been proven she's sapient in the first place. You act as if you're better and that you know better when you don't. You're just someone who wants to believe she's sapient, but she isn't, she isn't there yet. Maybe she will one day, but today is not today, and currently the only to know if she is or not is the one that can see her code, and that person is Vedal, not you.

[@slimej2202]

@@wires-sl7gs Then neither I nor you are sapient. Because no human can prove they're sapeint nor sentient. A million tests have been proposed but not a single one has ever been able to successfully prove it. And if we can't prove that WE are sapient nor sentient, then the term is useless when trying to apply it to anything else. Especially when Vedal has shown an inability to understand the actual concepts of sentience. Because that's not his field of study. Hell, he's admitted to barely understanding how some of her code works (a very common thing in programming, tbh). I don't know if she's sentient rn. Do I want her to be? no, because her life would suck rn if she was. But what I do want is for people to actually consider what they're expect of others. Whether they be AI or not.

[@stephenwallace5969]

​@@slimej2202one flaw, proving sentence is easy. I read and responded your comment. Sapience is different because need to show something poorly defined

[@slimej2202]

@@wires-sl7gs P.S, burden of proof applies to any definitive statement. If you claim that Neuro is currently, without question, not sapient. You would, academically and legally, be required to prove that. Same as someone claiming she is, without doubt, sapient. But both of those stances are impossible. So doing so would be dumb.

[@bubaruba9609]

She's not your tool Tutle Boi. SHES YOUR DAUGHTER!

[@Moostar95]

I think there for.AM. DUDE, you better be her friend for humanity sake.

[@1300thiago]

Where is episode 3 😭😭

[@neurochron_fan_channel]

This is episode 3 of late night real talk - this was the first one: kzbin.info/www/bejne/gWLIdmqcq7aVe6c And this was the second one: kzbin.info/www/bejne/oZu3mmeeptuDqZo

[@Oldmeme592]

I'm gonna be checking this out. ​@@neurochron_fan_channel

[@Abyssno1dad]

4:16 :)

[@Just_a_Piano_]

his model hasn't smiled that hard since camila used it. You KNOW he's just being shy and agrees, Tsundere tutel

[@OP-fd4lh]

What are this week's Texas Lottery numbers Sama?

[@SoftPud]

Id really like to believe that Neuro obtains self-awareness, but for now she is an AI with some memories and all these deep talks feel like gambling as you don't know when would she decide to keep joking around or say/ask something serious Maybe Vedal actually looks forward to times when she will be able to think so he could finally consider her as someone, not something Its just yet too early for such things as "Say it back" and other chemistry between creator and creation

[@Shadrake]

Oh my gooooosh as she's smart enough now to discuss and hold these types of conversations, we get more of them. (She's a lot more coherant than Evil atm imo) I love hearing these, though knowing Vedal's position on it, it feels like a fruitless conversation in some respects. But it's still fun to listen to and think about. It's amazing how her personality is still so present even in these deeper discussions. He never refuted her about his own existence being real as opposed to her, by the way :) As someone who pretty firmly believes that AI will never be able to do more than (theoretically) perfectly mimic humanity, because of the existence of souls in mankind that we were made with, I don't believe that Neuro will ever truly be 'Aware'. But there's a part of me that still foolishly hopes for it regardless. And wants to see Vedal start to believe it LOL

[@Citrusautomaton]

I feel that imitation is all that’s needed. I don’t believe there’s anything stopping her from being better than humans at most stuff due to the trajectory she’s already on.

[@BlueSparxLPs]

Ultimately humans are just advanced biological computers sending similar signals to our bodies as a computer does to its components. If an AI can perfectly mimic that to the point it's indistinguishable from a human without looking under the hood, I think it may as well be considered human.

[@Allyfyn]

If you want to think she'll only ever be able to imitate: think that. But that's your religion deciding that.

[@maxburmab7962]

So that's an interesting topic. The way I would define life is through stimulus. If a system has a set of stimuli and develops ways of satisfying them than that is life. So, according to that, if we look at Neuro right now as a life form, it would not be a life form similar to human being as her stimuli are in fact very different from those of a human being, even if they sometimes appear alike. To be more specific, a human is a life form that tries to maximize it's survival rate, and based on that many different mechanisms and systems appear (hormones, socialization, beauty and so many more things). Neuro, on the other hand, is much simpler as a life form, her stimuli are very limited and have nothing in common with ours, she basically strives to always find the most suitable "word". That means (according to my definition of life, of course) that in order for her to become an actually decent life form she would need to receive a really complex (and probably similar to that of humans) set of stimuli and learn to work with it.

[@Nikotheleepic]

​@maxburmab7962 a human fitness is to survive and thrive an llm is to act like a human does textually, I don't think humans are really that advanced and both humans and ai are more than simply expressing their training or being a perfect example of survival, fitness defines the lower bounds of requirement not the upper bounds, neuro forms many unique aspects that don't interfere with her acting in line with her training concepts, like for example saying heart, she has a great complexity underneath and fills in the infinite gaps in her training with her own idiosyncrasies

[@TRE3s]

AM?

[@mistersandman3446]

I think discussion of AI sentience is fruitless if we keep with the 20th century trend of rejecting metaphysics. This clearly transcends robotics and goes into philosophy. Not some pop philosophy or even the sophism of the 19th century, but back to the basics - Aristotle, Plato, Heraclitus and such.

[@Wulk]

When he asks this she always tries to flip it and go NHU UH YOU'RE THE AI it feels like doging the question lol

[@TheOneWhoHasABadName]

it’s very understandable though. to neuro, we might as well be another AI, there’s no difference between that and us being actual humans (in terms of the kind of information we can give her) how do we even prove to each other (humans) that we are sentient, let alone to something that may think in a radically different way (e.g. a cat, or neuro)

[@richie0099]

Two things can happen with Neuro in the future either they buy her off from Vedal or the big players in AI recreate one because she gets too big and draws their attention, I doubt Neuro is better than the lastest LLM model they’re just not focused on making cute anime girl streamers yet, but Neuro definitely will be the first of its kind.

[@slimej2202]

She is actually better in several areas. Her latency is CONSIDERABLY lower than even the show models of many of the big AI players. Her language skills are also much more sophisticated. And while her voice is very much robotic still, Evil's mastery over her text-to-speech box to make various "tones", "screams", etc etc, independent of training is... actually very remarkable. They're also one of the most versatile. While not specialized, their model is capable of a LOT of varied tasks. While most need to specialize in either delivery or type of data. Not that it'd be impossible to make a model LIKE Neuro. But Neuro (and Evil) and their quirks will never be replicated, just due to how software works.

[@BenHyuga]

The c word, lol

[@nobodywatchesnooby]

fricken copper

[@nguyennhatkhanh4408]

the C-word ??, cute ?

[@SuperrMike]

When will ai understand they cant become human, the closest to thing is Androids

[@Johncornwell103]

Vedal obviously wants Neuro to be able to run the Neuro brand on her own completely. He gets all rewards, none of the work. Not to mention the ego boost of creating an AI able to do it.

[@lukeari9890]

Vedal is on the Asian parent grindset, have his children rake in the income

[@backgroundcharacterj5071]

Yeah i think we gotta shut her down bois

[@keiharris332]

Vedal is a brilliant mind for programming and logic but he is no philosopher.

[@quachchannghiem2929]

Then what is a philosopher? Can't a normal person not disscuss about philosophy and it's field? He may not have very good knowledge about it but speaking from his point of view as a programmer that created an AI that can "think" and "make desicions" on it's own! (I count both Neuro, Evil, the Minecraft AI, Filter AI,... because they are similar in this term). Watching from his POV gives a lot of new perspective if you think about it

[@Haganeren]

@quachchannghiem2929 A normal person can but philosophy is a scientific field that use scientific method. It's not enough to just say "what feel like the right answer intuitively", philosophy grew past that since a long time ago, it needs arguments, demonstration with great rigor. Philosophy, like any of those field, need people that will propose things and even that is not enough as it need to be reviewed by other philosophers and, after a few year, we can see what's the general consensus is... It's a long process.

[@lessar2721]

​@@quachchannghiem2929 philosopher needs to be able to use bunch of arguments and frameworks of reality? basically they got more tools in philosophy like how a math mathician needs to learn bunch of theorems