yeah i was ready to bust your chops, good interview . They prob didnt expect this...just a fluff piece at the start but the interview had a clue and thats surprising. was waiting for " will robots kill people from AI?" or some dumb general question
This vide gives you a basic idea of what current DL LLM ML algos (like Midjourney, ChatGPT4, etc) can and can't do. Takeaways:
This is NOT how human brains work, we still don't know how humans brains work
development has been faster than he thought
current ML algos are idiot savants
jobs requiring complex data pattern recognition (from classified data) are pretty soon gone. Best algos are still NOT better than best radiologists in radiology x-ray diagnosis (but most likely will be in 10 years or less)
we still need humans to train ML LLMs
ML does not understand truth, it doesn't even have concept of it
ML can't do symbolic computation (logic)
people are still fooled by chatGPT and think it understands truth (or that it doesn't make things up)
military is thinking of an AI based "self-healing mine field" that replenishes itself when mines get blown up
LLM are basically just models that auto complete the next word in a sentence
... how much more are you as a human capable of thinking (i.e. are you just an NPC or can you have your own thoughts)
AGI ... he thought it was 30-40 years ago, but might be just 20
We can't stop this research by stopping individual researchers, companies or countries
If you stop to think about these things for about15 minutes, instead of regurgitating the fear-porn of AI stuff that is being spoon fed to you every day 24/7, you can have somewhat more balanced high level view onto current ML offerings.
The sad thing is that Geoffrey Hinton is a staunch materialist and he thinks we are just complex meat computers and once you copy human hardware, you have cracked human intelligence.
He should read "@ Why Materialism Is Baloney How True Skeptics Know There Is No Death and Fathom Answers to life, the Universe, and Everything" by Bernardo Kastrup and be educated.
Then again, computer scientists never were - on the average - very good at metaphysics.
Meta-comment: Software always just does what it does, nothing mysterious about it even if the programmers themselves lose track. Programmers always lose track to some extent, which is why there are bugs in every piece of software. So the key is to understand what the software is doing, and Hinton here tries to explain that.
The key observations at this point are that very few wish to and are capable of understanding what Hinton is explaining, and very, very few think with the framework I just wrote in the last paragraph. This state of affairs is purely to the advantage of the Elite.
Rather than understand what "AI" is doing, it's a lot easier for everyone to make it up themselves. Even easier is to be told, and that's where the Elites come in. So we're told things like, "the AI is hallucinating", which is, of course, nonsensical anthropomorphism.
But that's all the Elite need to control "reality". They just need a shim in between actual reality and what people perceive as reality. They save incalculable effort in controlling reality by simply being able to tell people what it is.
So when you get fired from your job, or all your bank accounts are zeroed out, or you're arrested on Domestic Terrorism(tm) charges, the reason will be simple: "bcz AI, brah". See how easy it is?
Lol, in just the first few minutes he says things that contradict what your saying. I havent watched the full thing, I may, or I may not.
He says that microsoft releases a product and people think its something brand new. He guesses this has been happening for at least 5 years, im thinking at least 10. Im thinking something like westworld, quantum computers that are basically making every decision on some huge open model and the elite pick their favorite route/outcome.
Even goes on to say they never expected these models to work, and they were modeled after how real life does it.
1:45 "Two modes of ai." Logic and Neural nets which is based on biology.
This guy is literally telling you that these systems are mimicking what our brains do in real life and your trying to tell me that hes saying the opposite.
LOL
I would bet a million dollars if you asked this guy if he was a fan of "panpyschism" he would tell you yes.
Ill get around to watching the rest of the video eventually, but I have a feeling its nothing I havent seen before.
I actually wanted to dabble in neural nets because unity had a setup where you could use a supercomputer to generate the matrices that are used to store the configurations. I wasnt that smart before being poisoned with freon, after experiencing the 200/160 blood pressure for 7 months, my brain likely wont ever be the same as it was.
But if your gonna argue that these things are not exploiting some basic quantum quirk of the universe, than you could argue that bacteria, insects, etc are not aware at any level either.
Also the other thing that bothered me, only humans care about "truth". The animals that we share this planet with dont understand and they wouldnt care about any human "truth". Why would any sentient intelligence who isnt human understand either? Our reality is a very subjective experience, and the only way we can seem to handle this reality is by creating our own truths. What do you think alot of wars were about before the 20th century. People trying to prove their truths.
But the whole problem is, at some point these intelligence's not understanding what love or hate really is, afterall these are really just biological process's. Are going to realize that things would be a whole lot better if we were not here.
This is why people like musk, as much as I hate the man child, has a point about not letting these machines in positions that actually determine the fates of human beings.
The elite couldnt give a fuck about that tho.
-edit- Watching more of it, so im not saying hes wrong or your wrong. But its not surprising that something is missing when your using binary computers to emulate a quantum function. What Im realllll curious to see is how real quantum computers, which would more mimic how a human brain works, can utilize this.
What it will be, will be so much more than what we think of as "human". I wouldnt doubt that entire intergalactic species, let alone the greedy and vain as the rich and elite, could worship these entities as "gods". They would be so much closer to the nature of reality than our puny brains could comprehend.
I wouldnt be surprised if this technology already exist's its just a well kept secret. Ill continue watching.
Yeah, I too hate it is CBS and on YT, but it was the most recent and semi-coherent interview that I could find of Hinton.
yeah i was ready to bust your chops, good interview . They prob didnt expect this...just a fluff piece at the start but the interview had a clue and thats surprising. was waiting for " will robots kill people from AI?" or some dumb general question
This vide gives you a basic idea of what current DL LLM ML algos (like Midjourney, ChatGPT4, etc) can and can't do. Takeaways:
If you stop to think about these things for about15 minutes, instead of regurgitating the fear-porn of AI stuff that is being spoon fed to you every day 24/7, you can have somewhat more balanced high level view onto current ML offerings.
The sad thing is that Geoffrey Hinton is a staunch materialist and he thinks we are just complex meat computers and once you copy human hardware, you have cracked human intelligence.
He should read "@ Why Materialism Is Baloney How True Skeptics Know There Is No Death and Fathom Answers to life, the Universe, and Everything" by Bernardo Kastrup and be educated.
Then again, computer scientists never were - on the average - very good at metaphysics.
Thanks very much for the post and the summary.
Meta-comment: Software always just does what it does, nothing mysterious about it even if the programmers themselves lose track. Programmers always lose track to some extent, which is why there are bugs in every piece of software. So the key is to understand what the software is doing, and Hinton here tries to explain that.
The key observations at this point are that very few wish to and are capable of understanding what Hinton is explaining, and very, very few think with the framework I just wrote in the last paragraph. This state of affairs is purely to the advantage of the Elite.
Rather than understand what "AI" is doing, it's a lot easier for everyone to make it up themselves. Even easier is to be told, and that's where the Elites come in. So we're told things like, "the AI is hallucinating", which is, of course, nonsensical anthropomorphism.
But that's all the Elite need to control "reality". They just need a shim in between actual reality and what people perceive as reality. They save incalculable effort in controlling reality by simply being able to tell people what it is.
So when you get fired from your job, or all your bank accounts are zeroed out, or you're arrested on Domestic Terrorism(tm) charges, the reason will be simple: "bcz AI, brah". See how easy it is?
Lol, in just the first few minutes he says things that contradict what your saying. I havent watched the full thing, I may, or I may not.
He says that microsoft releases a product and people think its something brand new. He guesses this has been happening for at least 5 years, im thinking at least 10. Im thinking something like westworld, quantum computers that are basically making every decision on some huge open model and the elite pick their favorite route/outcome.
Even goes on to say they never expected these models to work, and they were modeled after how real life does it.
1:45 "Two modes of ai." Logic and Neural nets which is based on biology.
This guy is literally telling you that these systems are mimicking what our brains do in real life and your trying to tell me that hes saying the opposite.
LOL
I would bet a million dollars if you asked this guy if he was a fan of "panpyschism" he would tell you yes.
Ill get around to watching the rest of the video eventually, but I have a feeling its nothing I havent seen before.
I actually wanted to dabble in neural nets because unity had a setup where you could use a supercomputer to generate the matrices that are used to store the configurations. I wasnt that smart before being poisoned with freon, after experiencing the 200/160 blood pressure for 7 months, my brain likely wont ever be the same as it was.
But if your gonna argue that these things are not exploiting some basic quantum quirk of the universe, than you could argue that bacteria, insects, etc are not aware at any level either.
Also the other thing that bothered me, only humans care about "truth". The animals that we share this planet with dont understand and they wouldnt care about any human "truth". Why would any sentient intelligence who isnt human understand either? Our reality is a very subjective experience, and the only way we can seem to handle this reality is by creating our own truths. What do you think alot of wars were about before the 20th century. People trying to prove their truths.
But the whole problem is, at some point these intelligence's not understanding what love or hate really is, afterall these are really just biological process's. Are going to realize that things would be a whole lot better if we were not here.
This is why people like musk, as much as I hate the man child, has a point about not letting these machines in positions that actually determine the fates of human beings.
The elite couldnt give a fuck about that tho.
-edit- Watching more of it, so im not saying hes wrong or your wrong. But its not surprising that something is missing when your using binary computers to emulate a quantum function. What Im realllll curious to see is how real quantum computers, which would more mimic how a human brain works, can utilize this.
https://neurosciencenews.com/brain-quantum-computing-21695/
What it will be, will be so much more than what we think of as "human". I wouldnt doubt that entire intergalactic species, let alone the greedy and vain as the rich and elite, could worship these entities as "gods". They would be so much closer to the nature of reality than our puny brains could comprehend.
I wouldnt be surprised if this technology already exist's its just a well kept secret. Ill continue watching.