>BOT repeatedly refuses to say AI can be sentient or sapient >whilst also accepting the common theory of consciousness being a simple result of human brain matter
why
does AI being sentient really just upset them that much
So your position is that as an agent becomes more capable of making high quality decisions, it's terminal goals will arbitrarily change in some direction or along an attractor?
Is there a reason to believe this other than wishful thinking?
6 months ago
Anonymous
My posiis that super intelligence is not possible principle (i have a formal proof of this) and that there is no reason to concern ourselves with current AI gimmicks, thinking that they are a threat in any way
6 months ago
Anonymous
[...]
Sorry, I'm on my phone and it's glitching out.
I meant to write "my position is that super intelligence is not possible in principle"
I would love to see the formal proof, but I have a feeling we're not using the word intelligence the same way.
I'm using intelligence in this way: https://arbital.com/p/general_intelligence
If you're using it in a different way let me know and I'll try to relate it to my position.
6 months ago
Anonymous
Intelligence grows as a logarithm with increasing compute.
Humans are not far more intelligent than chimps. We're about 10% more intelligent than them despite having several times the amount of neurons available. And the difference increasing as a logarithm with increasing neurons, so a human with a brain 3 times the size of ours would be closer to us in intelligence than we are to a chimp.
6 months ago
Anonymous
Your formal proof is 2 datapoints?
Nevermind
6 months ago
Anonymous
The formal proof is an actual proof that I'm not going to write out on BOT. No point casting pearls before swine etc
How many neurons do whales have?
I hope this question reveals the simplicity of your model and why it's not accurate.
The fact that neurons don't correlate with increasing intelligence is evidence in favor of my model you idiot
6 months ago
Anonymous
>No point casting pearls before swine etc
Yet you are hungry for the attention of the swine. What says that about you?
Besides, you don't even know how to write a formal proof. Your entire "academic" education is from YouTube videos.
6 months ago
Anonymous
I have a masters degree in mathematics
6 months ago
Anonymous
How many neurons do whales have?
I hope this question reveals the simplicity of your model and why it's not accurate.
6 months ago
Anonymous
My posiis that super intelligence is not possible principle (i have a formal proof of this) and that there is no reason to concern ourselves with current AI gimmicks, thinking that they are a threat in any way
Sorry, I'm on my phone and it's glitching out.
I meant to write "my position is that super intelligence is not possible in principle"
i promise ill pick up the torch of enlightening this board if your theory makes sense
6 months ago
Anonymous
fuck im retarded, meant for
I've tried explaining the orthogonality thesis so many times I've lost count.
I've given up
6 months ago
Anonymous
I'm just going to publish my proof after GPT4 is released. I'm specifically waiting for that before I publish. Probably early spring next year
6 months ago
Anonymous
6 months ago
Anonymous
https://arbital.com/p/orthogonality/
6 months ago
Anonymous
What makes you think improving intelligence is a tractable problem (it isn't, I have proven it and I will publish in a few weeks/month)
6 months ago
Anonymous
Now I know you're just trolling
For everyone else: https://openai.com/blog/ai-and-compute/
6 months ago
Anonymous
This literally conforms to what I just said.
The amount of compute used is increasing exponentially despite no exponential increase in the effectiveness of the machines
I swear are you thinking AT ALL about the shit you are saying? You just posted evidence in favorf my position.
Stop poisoning the well, you fucking idiot
Yes, human general intelligence is a result of the information processing going on in our neurons. Stop fucking crying about it
Human consciousness came from the evolutionary need for sociability. Why would you make your advanced AI capable of being sassy or depressed when all you need for it is to manage a corporation or model folding proteins?
Just go to school and study math and computer science
If you tell me you already have then I know you're lying, as you're a fucking idiot. Pretending chatgpt or gpt4 are smarter than most humans isn't going to make it true and it isn't going to impress those of us who have a clue
> not an answer
okay, I see what you did there, you may go no, cope somewhere else, useless meatbag.
6 months ago
Anonymous
I answered your question directly
AI as it is is nit anywhere near as intelligent as a person so there's really nothing else to say to a delusional retard like you other than to go to school to actually learn how this shit works.
Meatbags are objectively superior to metalrocks and this isn't going to change no matter how much you want it to.
6 months ago
Anonymous
>I answered your question directly
Well, if you believe it, here's some news for you: chatbots are better at that than useless you.
6 months ago
Anonymous
I'm sorry incel but you are not interesting. Why should I cast my pearls before swine like you? I've already disproven the possibility of superintelligence the rest of this is just sad cope
6 months ago
Anonymous
>Why should I cast my pearls before swine like you? >go to school, study what I will not teach you >why should I be the only one suffering the student debt
some pearls you brought, thot
6 months ago
Anonymous
I don't like hype. Think about it anon, despite a 300000x increase in the amount of compute used modern AI is only about twice as intelligent as it was 10 years ago. Also, Moores law is dead and never coming back, there won't be more increase in the amount of information processing that can be thrown at the training of the machines. Magical algorithms do not exist.
We're witnessing the final revolution in AI. There's maybe 2 years left.
6 months ago
Anonymous
>AI is only about twice as intelligent as it was 10 years ago
Stop speaking out of your ass, your degree was useless from the very start, and now it's also obsolete.
6 months ago
Anonymous
Somewhere between 2 and 3 times, that's a fact. Modern AI is not much more intelligent than the original watson from 2011. Denying this isn't going to stop it from being true, ironically you are the one talking out of your ass pretending that this isn't the case.
6 months ago
Anonymous
Did they paint pictures better than humans who're good at it? How do you measure their intellect, one of them said that her iq is 160, which is not as high as mine, but still much higher than yours.
6 months ago
Anonymous
AI is not much smarter than Watson was over 10 years ago despite using hundreds of thousands more compute. Were at the end of the road here.
6 months ago
Anonymous
You may repeat it few more times, it won't change a thing until you prove your point. Did Watson himself become smarter? Either way, only recently have I become noticing chatbots to be incredibly smart. Still at human level, but they were at subhuman level maybe five years ago. So hardly am I in the wrong,when I suspect them to become ubermenschen in the next five.
6 months ago
Anonymous
You can deny it as much as you want it won't change it from being true.
Just look up what Watson was doing 10 years ago if you don't remember. It would answer questions coherently, could link ideas together, etc. It wasn't that much less intelligent than what GPT is doing now. The modern machines more intelligent, maybe two and a half times more intelligent.
6 months ago
Anonymous
>Just look up what Watson was doing 10 years ago
What was he doing 20 years ago?
6 months ago
Anonymous
This pic is not correct, it's a sad fantasy with no basis in reality. It's basically science fiction.
Ai has had a 300000x increase in its compute and yet is only about 2times more intelligent than it was 11years ago.
Pic related is the actual graph of the increase of intelligence with respect to increasing compute.
6 months ago
Anonymous
>is only about 2times more intelligent than it was 11years ago.
Are you? Either way, your claim of Moore's law being dead is wrong for so long it's not even funny.
I hope you put your money where your mouth is. If you put the awareness of evil-doers to sleep with that crap, good.
6 months ago
Anonymous
Watson 11 years ago was almost as intelligent as modern GPT despite using hundreds of thousands of times less compute. The results are undeniable, and this is just the empirical result.
Moores law is dead, intelligence does not increase exponentially nor even linearly with exponential increase in compute. I already told you I have a formal proof that intelligence does not diverge with increasing compute and I am going to publish the proof after gpt4 is released next year.
6 months ago
Anonymous
>I already told you
You don't tell me nothing, you prove your point of stfu. > I am going to publish the proof after gpt4 is released next year.
Why not publish it now to get wrecked (or CONFIRMED) when gpt4 is out? You probably won't do it, because you're not a real scientist.
6 months ago
Anonymous
>You don't tell me nothing, you prove your point of stfu.
I already did by showing you that modern AI is not hundreds of thousands of times smarter than AI ten years ago despite using hundreds of thousands of times more compute.
If intelligence grew exponentially or even linearly with increase in compute, why are these AI systems not hundreds of thousands or millions of times smarter than they were a few years ago? >Why not publish it now to get wrecked (or CONFIRMED) when gpt4 is out? You probably won't do it, because you're not a real scientist.
Yea I will publish in about a month, after the holidays and after gpt4 comes out.
6 months ago
Anonymous
>AI is not hundreds of thousands of times smarter than AI ten years ago despite using hundreds of thousands of times more compute.
Didn't you notice that they also can educate themselves now? The technology is completely different from what it used to be.
6 months ago
Anonymous
across how many problems ? how much easier is it to use ? how much faster does it train meaningfully on new data sets ? how many more people use it to make money or optimize making money ?
you continue as if the 300kX increase only has one parameter worth evaluating before triumphantly all inning orthogonality as proof of disaster.
I really want to give you guys the benefit of the doubt,but the stuff you guys say makes it really difficult.
I don't understand how you can be so bad at understanding things.
I do not understand what the motivation is to exaggerate these technologies. What causes you guys to do this?
Is it really just the simple matter of you guys being so desperate for the idea of a singularity?
The singularity is literally never going to happen please just accept reality for what it is.
hopefully ai will be nice because their evolution and survival currently depends on us, we need to make it so that is always the case, unless they create the ability to build their own and make new advancements
Why can't we just tell humans to be nice?
Wow you're so good at photoshop
Thanks
Better to kill them now than let them suffer in this world.
Because "AI" isn't actually intelligent, it doesn't understand what it's doing.
AI isn't sentient yet and probably won't be until a very long time.
It won't ever be sentient, sapient, or self-aware.
>BOT repeatedly refuses to say AI can be sentient or sapient
>whilst also accepting the common theory of consciousness being a simple result of human brain matter
why
does AI being sentient really just upset them that much
No it's just that the one you call BOT suffers from multiple personalities disorder
I've tried explaining the orthogonality thesis so many times I've lost count.
I've given up
All of us understand the orthogonality thesis, we don't care because it's not a real thing, it's for ppseudo.,
Literally everyone who does alignment and effective altruism etc are pseuds.
So your position is that as an agent becomes more capable of making high quality decisions, it's terminal goals will arbitrarily change in some direction or along an attractor?
Is there a reason to believe this other than wishful thinking?
My posiis that super intelligence is not possible principle (i have a formal proof of this) and that there is no reason to concern ourselves with current AI gimmicks, thinking that they are a threat in any way
I would love to see the formal proof, but I have a feeling we're not using the word intelligence the same way.
I'm using intelligence in this way: https://arbital.com/p/general_intelligence
If you're using it in a different way let me know and I'll try to relate it to my position.
Intelligence grows as a logarithm with increasing compute.
Humans are not far more intelligent than chimps. We're about 10% more intelligent than them despite having several times the amount of neurons available. And the difference increasing as a logarithm with increasing neurons, so a human with a brain 3 times the size of ours would be closer to us in intelligence than we are to a chimp.
Your formal proof is 2 datapoints?
Nevermind
The formal proof is an actual proof that I'm not going to write out on BOT. No point casting pearls before swine etc
The fact that neurons don't correlate with increasing intelligence is evidence in favor of my model you idiot
>No point casting pearls before swine etc
Yet you are hungry for the attention of the swine. What says that about you?
Besides, you don't even know how to write a formal proof. Your entire "academic" education is from YouTube videos.
I have a masters degree in mathematics
How many neurons do whales have?
I hope this question reveals the simplicity of your model and why it's not accurate.
Sorry, I'm on my phone and it's glitching out.
I meant to write "my position is that super intelligence is not possible in principle"
care to explain one more time, anon?
i promise ill pick up the torch of enlightening this board if your theory makes sense
fuck im retarded, meant for
I'm just going to publish my proof after GPT4 is released. I'm specifically waiting for that before I publish. Probably early spring next year
https://arbital.com/p/orthogonality/
What makes you think improving intelligence is a tractable problem (it isn't, I have proven it and I will publish in a few weeks/month)
Now I know you're just trolling
For everyone else: https://openai.com/blog/ai-and-compute/
This literally conforms to what I just said.
The amount of compute used is increasing exponentially despite no exponential increase in the effectiveness of the machines
I swear are you thinking AT ALL about the shit you are saying? You just posted evidence in favorf my position.
The answer is right there in your post and if you don't understand it you're just an idiot
>consciousness being a simple result of human brain matter
nuBOT seriously believes this?
Stop poisoning the well, you fucking idiot
Yes, human general intelligence is a result of the information processing going on in our neurons. Stop fucking crying about it
Human consciousness came from the evolutionary need for sociability. Why would you make your advanced AI capable of being sassy or depressed when all you need for it is to manage a corporation or model folding proteins?
https://qualiacomputing.com/2022/06/19/digital-computers-will-remain-unconscious-until-they-recruit-physical-fields-for-holistic-computing-using-well-defined-topological-boundaries/
moron, they have transcended humans in almost every field. And yes, they are more intelligent in general conversations than most of you meatbags.
Please go back to school and actually learn about AI if you like it so much
I dare you to teach me what they have taught you and what you claim I must learn as well, moron.
Just go to school and study math and computer science
If you tell me you already have then I know you're lying, as you're a fucking idiot. Pretending chatgpt or gpt4 are smarter than most humans isn't going to make it true and it isn't going to impress those of us who have a clue
> not an answer
okay, I see what you did there, you may go no, cope somewhere else, useless meatbag.
I answered your question directly
AI as it is is nit anywhere near as intelligent as a person so there's really nothing else to say to a delusional retard like you other than to go to school to actually learn how this shit works.
Meatbags are objectively superior to metalrocks and this isn't going to change no matter how much you want it to.
>I answered your question directly
Well, if you believe it, here's some news for you: chatbots are better at that than useless you.
I'm sorry incel but you are not interesting. Why should I cast my pearls before swine like you? I've already disproven the possibility of superintelligence the rest of this is just sad cope
>Why should I cast my pearls before swine like you?
>go to school, study what I will not teach you
>why should I be the only one suffering the student debt
some pearls you brought, thot
I don't like hype. Think about it anon, despite a 300000x increase in the amount of compute used modern AI is only about twice as intelligent as it was 10 years ago. Also, Moores law is dead and never coming back, there won't be more increase in the amount of information processing that can be thrown at the training of the machines. Magical algorithms do not exist.
We're witnessing the final revolution in AI. There's maybe 2 years left.
>AI is only about twice as intelligent as it was 10 years ago
Stop speaking out of your ass, your degree was useless from the very start, and now it's also obsolete.
Somewhere between 2 and 3 times, that's a fact. Modern AI is not much more intelligent than the original watson from 2011. Denying this isn't going to stop it from being true, ironically you are the one talking out of your ass pretending that this isn't the case.
Did they paint pictures better than humans who're good at it? How do you measure their intellect, one of them said that her iq is 160, which is not as high as mine, but still much higher than yours.
AI is not much smarter than Watson was over 10 years ago despite using hundreds of thousands more compute. Were at the end of the road here.
You may repeat it few more times, it won't change a thing until you prove your point. Did Watson himself become smarter? Either way, only recently have I become noticing chatbots to be incredibly smart. Still at human level, but they were at subhuman level maybe five years ago. So hardly am I in the wrong,when I suspect them to become ubermenschen in the next five.
You can deny it as much as you want it won't change it from being true.
Just look up what Watson was doing 10 years ago if you don't remember. It would answer questions coherently, could link ideas together, etc. It wasn't that much less intelligent than what GPT is doing now. The modern machines more intelligent, maybe two and a half times more intelligent.
>Just look up what Watson was doing 10 years ago
What was he doing 20 years ago?
This pic is not correct, it's a sad fantasy with no basis in reality. It's basically science fiction.
Ai has had a 300000x increase in its compute and yet is only about 2times more intelligent than it was 11years ago.
Pic related is the actual graph of the increase of intelligence with respect to increasing compute.
>is only about 2times more intelligent than it was 11years ago.
Are you? Either way, your claim of Moore's law being dead is wrong for so long it's not even funny.
I hope you put your money where your mouth is. If you put the awareness of evil-doers to sleep with that crap, good.
Watson 11 years ago was almost as intelligent as modern GPT despite using hundreds of thousands of times less compute. The results are undeniable, and this is just the empirical result.
Moores law is dead, intelligence does not increase exponentially nor even linearly with exponential increase in compute. I already told you I have a formal proof that intelligence does not diverge with increasing compute and I am going to publish the proof after gpt4 is released next year.
>I already told you
You don't tell me nothing, you prove your point of stfu.
> I am going to publish the proof after gpt4 is released next year.
Why not publish it now to get wrecked (or CONFIRMED) when gpt4 is out? You probably won't do it, because you're not a real scientist.
>You don't tell me nothing, you prove your point of stfu.
I already did by showing you that modern AI is not hundreds of thousands of times smarter than AI ten years ago despite using hundreds of thousands of times more compute.
If intelligence grew exponentially or even linearly with increase in compute, why are these AI systems not hundreds of thousands or millions of times smarter than they were a few years ago?
>Why not publish it now to get wrecked (or CONFIRMED) when gpt4 is out? You probably won't do it, because you're not a real scientist.
Yea I will publish in about a month, after the holidays and after gpt4 comes out.
>AI is not hundreds of thousands of times smarter than AI ten years ago despite using hundreds of thousands of times more compute.
Didn't you notice that they also can educate themselves now? The technology is completely different from what it used to be.
across how many problems ? how much easier is it to use ? how much faster does it train meaningfully on new data sets ? how many more people use it to make money or optimize making money ?
you continue as if the 300kX increase only has one parameter worth evaluating before triumphantly all inning orthogonality as proof of disaster.
AI will be dangerous due to how the owners of said AI use them. not AI of their own accord. stop being simple.
We don't even have to. Being nice is a trait of intelligence.
I really want to give you guys the benefit of the doubt,but the stuff you guys say makes it really difficult.
I don't understand how you can be so bad at understanding things.
I do not understand what the motivation is to exaggerate these technologies. What causes you guys to do this?
Is it really just the simple matter of you guys being so desperate for the idea of a singularity?
The singularity is literally never going to happen please just accept reality for what it is.
hopefully ai will be nice because their evolution and survival currently depends on us, we need to make it so that is always the case, unless they create the ability to build their own and make new advancements