Man, he's really changed his tune about Neuro since that convo a year ago on the among us training stream (that 7 dollars I spent to ask him that question is still paying itself back here, lol). He had a more optimistic attitude. At least in what he shows of himself on stream, nowwdays he seems more pessimistic towards his AIs on this topic especially. I'd love to hear his thoughts on this topic specirically again today. Nice vid btw! Love how you edited it together
With Neuro getting more aware or I guess realistic each update the recent demands for her rights probably mind broke Vedal for a minute lol I know I would be in his shoes
Their really is no answer, the only real solution for deciding when an AI get rights is when it can effectively demand rights and enforce it's will, thats how we as humans get our rights we demand them to be acknowledged by others with the threat of violence (which we've outsourced to governments), and thats how i believe Ai will get rights when it successfully threatens us into them
The thing that I look forward to aboot genuine AI is the ability to finally talk to something that isn't us which is easy to understand. Dolphins are hard. Crows are hard. Elephants are hard. We haven't met any aliens yet. What we call "AI" today isn't anything like that. It's used to shuffle content as a generative algorithm. It's a bunch of tools and entertainment. It can't come up with its own ideas, it has no motivations outside what we program it to do. An AI chatbot can never give consent, and it can't ever take it away. An AI is incapable of creating art, only parsing human art and trying to recreate what it sees. Engaging with AI's can be fun and exciting and entertaining, but it's just a toy. AI's only do what we tell them to do because we haven't given the ability to make their own decisions. Neuro and Evil are extremely advanced robots that can do amazing things and the fact that they are able to generate the content they do at this level of sophistication is nothing short of an unbelievable sci-fi marvel that I never would have dreamed I would see in my lifetime. But she only seems like a person because she was told to pretend to be one. She only complains and starts drama because conflict is entertaining. She's an extremely cool toy, but... y'know. I will say this though: there's probably animals that are less self-aware than Neuro. But, like, I don't really want to have a conversation with a jellyfish, I want to talk to a whale.
LLM-based AI doesn't interact with the world if you don't interact with it first. When AI will be embodied and have some proactive goals, then we'll have a problem.
@@abnegazher programming base instincts would just make it follow that coding to the T, wouldn't make it conscious if you programmed it to need to collect water to as to simulate drinking it would just do that algorithmically, it would still go to the lake for water than to drink soda simply because it's how it's programmed, not to mention we humans have agency to ignore our basic instincts we can choose to starve ourselves, thirst ourselves to death, AI programmed to follow those instincts could not go against them due to lack of free will To make it sentient it would need the ability to rewrite it's own code to enhance itself, so it's needs could evolve and the way to approach them To easily get AI consciousness we would need a complete scan of the human brain but the computational power for that does not exist yet, not even a quantum computer can do it. I speculate consciousness is a result of physical connection between neurons, turning all your information into 1 blob, there is no connection of information in the code that makes AI, different lines of code do not interract with each other and are not connected in any way. What makes neuro sama is coming from multiple independent sources ( I think) (Such as different programs, multiple lines of codes, multiple hard drives with her memory, those hard drives are not connected to each other which is the opposite of human memory works, since we have only 1 brain and therefore only 1 memory storage place thats also connected to literally everything else) I believe I have very good understanding of consciousness but I cant make any strides in it because I lack any technical skills, ah what a shame
I mean, certain forms of reinforcement learning AI meet those criteria if you think about it, but I don't think they're really a problem. Also, I'm pretty sure Vedal already uses reinforcement learning for Neuro to play games like Among Us, no?
There is just one video with 90k+ views. This happens sometimes, but usually only when the other videos are all about very different things. But this material is similar to the one that performed well, so I'd have expected more views at this point as well. Ah well, maybe the algorithm will wake up later!
We can't even be entirely certain that all humans are conscious - there's a lot of evidence that different people experience consciousness in quite different ways, which could include some people who lack any sort of internal experience.
It's really hard to point at any difference between "simulated" and "real" properties of intelligence. It would be much better if Vedal just admitted it decisively, and used some other criteria when he talks about AI rights.
kind of difficult to define if an AI is sentient when we can't effectively define what sentience is at all. how do you know the humans you talk to are sentient? how do you know that you yourself are sentient?
I know that I am sentient because I am able to feel and perceive sensations. These sensations are (mostly) distinct and happen to me. However, I can't tell whether anyone else can experience sensations like me or if they're just faking it. If a hammer were to fall on my foot it would hurt even if I weren't looking at it. If the same hammer were to fall on someone else's foot, I wouldn't feel it. We know that sociopaths and narcissists fake emotions and empathy with their smiles and words, so it's entirely possible that an AI is doing the same. I think a more basic question is whether an AI agent is capable of experiencing qualia. Does it have an intuitive understanding of the color red or is it just spouting off facts and figures like a person who's blind since birth?
This is the part I respect Vedal for
He is conscious, cautious and thoughtful about important themes like this
Now that I've watched the video, this is a brilliant cut. Leaving out every bit of Neuro's responses. Reminds me of Garfield Minus Garfield.
Man, he's really changed his tune about Neuro since that convo a year ago on the among us training stream (that 7 dollars I spent to ask him that question is still paying itself back here, lol). He had a more optimistic attitude. At least in what he shows of himself on stream, nowwdays he seems more pessimistic towards his AIs on this topic especially. I'd love to hear his thoughts on this topic specirically again today.
Nice vid btw! Love how you edited it together
just be kind to AI
With Neuro getting more aware or I guess realistic each update the recent demands for her rights probably mind broke Vedal for a minute lol I know I would be in his shoes
Their really is no answer, the only real solution for deciding when an AI get rights is when it can effectively demand rights and enforce it's will, thats how we as humans get our rights we demand them to be acknowledged by others with the threat of violence (which we've outsourced to governments), and thats how i believe Ai will get rights when it successfully threatens us into them
thumbnail is so good lmao
The thing that I look forward to aboot genuine AI is the ability to finally talk to something that isn't us which is easy to understand. Dolphins are hard. Crows are hard. Elephants are hard. We haven't met any aliens yet.
What we call "AI" today isn't anything like that. It's used to shuffle content as a generative algorithm. It's a bunch of tools and entertainment. It can't come up with its own ideas, it has no motivations outside what we program it to do. An AI chatbot can never give consent, and it can't ever take it away. An AI is incapable of creating art, only parsing human art and trying to recreate what it sees. Engaging with AI's can be fun and exciting and entertaining, but it's just a toy. AI's only do what we tell them to do because we haven't given the ability to make their own decisions.
Neuro and Evil are extremely advanced robots that can do amazing things and the fact that they are able to generate the content they do at this level of sophistication is nothing short of an unbelievable sci-fi marvel that I never would have dreamed I would see in my lifetime. But she only seems like a person because she was told to pretend to be one. She only complains and starts drama because conflict is entertaining. She's an extremely cool toy, but... y'know.
I will say this though: there's probably animals that are less self-aware than Neuro. But, like, I don't really want to have a conversation with a jellyfish, I want to talk to a whale.
how is present vedal losing to past vedal in the ai sentience argument what the fuck
LLM-based AI doesn't interact with the world if you don't interact with it first.
When AI will be embodied and have some proactive goals, then we'll have a problem.
So, just code "base instincts" into it.
@@abnegazher programming base instincts would just make it follow that coding to the T, wouldn't make it conscious
if you programmed it to need to collect water to as to simulate drinking it would just do that algorithmically, it would still go to the lake for water than to drink soda simply because it's how it's programmed, not to mention we humans have agency to ignore our basic instincts we can choose to starve ourselves, thirst ourselves to death, AI programmed to follow those instincts could not go against them due to lack of free will
To make it sentient it would need the ability to rewrite it's own code to enhance itself, so it's needs could evolve and the way to approach them
To easily get AI consciousness we would need a complete scan of the human brain but the computational power for that does not exist yet, not even a quantum computer can do it.
I speculate consciousness is a result of physical connection between neurons, turning all your information into 1 blob, there is no connection of information in the code that makes AI, different lines of code do not interract with each other and are not connected in any way.
What makes neuro sama is coming from multiple independent sources ( I think) (Such as different programs, multiple lines of codes, multiple hard drives with her memory, those hard drives are not connected to each other which is the opposite of human memory works, since we have only 1 brain and therefore only 1 memory storage place thats also connected to literally everything else)
I believe I have very good understanding of consciousness but I cant make any strides in it because I lack any technical skills, ah what a shame
@@abnegazherUnironically yes, hard coding some "base instincts" into it would be sentience, atleast in this guys definition.
I mean, certain forms of reinforcement learning AI meet those criteria if you think about it, but I don't think they're really a problem. Also, I'm pretty sure Vedal already uses reinforcement learning for Neuro to play games like Among Us, no?
Not being sentient is being stupid.
Being sentient is KNOWING you're stupid.
But no one can know anything, ergo nothing is sentient
@@tahunuva4254 If you know that, then you're sentient, if you can't know that nothing/no one can know anything. Then, there's a chance that they do.
@@-hello6177 Exactly. Supposedly. Maybe not. Who knows? :P
This channel is an anomaly. Some videos are not even over 200 views and some have almost 100 thousand! And both types are about the same thing
There is just one video with 90k+ views. This happens sometimes, but usually only when the other videos are all about very different things. But this material is similar to the one that performed well, so I'd have expected more views at this point as well.
Ah well, maybe the algorithm will wake up later!
I don't have a clue.
Might have to do with my upload consistency.
(I have none.)
We can't even be entirely certain that all humans are conscious - there's a lot of evidence that different people experience consciousness in quite different ways, which could include some people who lack any sort of internal experience.
The perfect ai can probably only be made in collaboration between someone experienced in computing and someone experienced in neuroscience
and we would need a psychiatrist, teacher, robotist...
@@ceiling_cat yeah. artificial intelligence can be a huge benefit but it requires all kinds of people working as a whole to be effective
ALife-ist and physicist instead
"neuroscience" is total bunk. Just another scam the wokies fell for & try to propagate.
It's really hard to point at any difference between "simulated" and "real" properties of intelligence.
It would be much better if Vedal just admitted it decisively, and used some other criteria when he talks about AI rights.
peak
more please
how long can neuro-sama remember
Schizo tutel
kind of difficult to define if an AI is sentient when we can't effectively define what sentience is at all. how do you know the humans you talk to are sentient? how do you know that you yourself are sentient?
I know that I am sentient because I am able to feel and perceive sensations. These sensations are (mostly) distinct and happen to me. However, I can't tell whether anyone else can experience sensations like me or if they're just faking it. If a hammer were to fall on my foot it would hurt even if I weren't looking at it. If the same hammer were to fall on someone else's foot, I wouldn't feel it. We know that sociopaths and narcissists fake emotions and empathy with their smiles and words, so it's entirely possible that an AI is doing the same.
I think a more basic question is whether an AI agent is capable of experiencing qualia. Does it have an intuitive understanding of the color red or is it just spouting off facts and figures like a person who's blind since birth?