The current LLMs and future ones on the same architecture won't cut it though. This is all about the hypothetical super powerful AI of the future. Much more research will be needed to get there.
Is it though? You only need someone to program self reprlication and survival into its augmentation and you're on your way to a run off. There also it's comprehension of logic.
@@devilsolution9781 I would not lose sleep over it for the coming years. Of course, we could connect AI-based systems to the real world today and wreak havoc beyond our imagination. But that will be a human error. The AI has no clue what it is doing.
Do you have more hope from regular people reacting and influencing governments, or from authorities taking the initiative regardless, cooperating, and eventually making the correct moves?
The power of manipulation is about more than intelligence as is power itself. The world would be better if it weren't, just think about the Manhattan Project and how few people understood what man hath wrought. But I'd never underestimate science's ability to expand misery across all sentience. "Science, you whore! You've done it again." -Dan Halen
The title doesn't match the video content. Also, Eliezer doesn't like to be called "AI researcher". IIRC, he asked to be introduced as a "decision theorist".
A lot of value statements. It's so powerful and so immense and dangerous and... powerful. AI smart. Manipulation. Technical pathways for any of those claims to be measured and that is something for an expert to discuss. One important thing is that it doesn't require emotion in order to maneuver creative intelligence towards an objective. This is a nice way of preventing irrational behavior and making it easier to self regulate. I'll run a subjective example by saying you can make an AI willing to kill itself and the whole notion that it's out of control spontaneously mimicking the complexity of human psychology without cascade failure.. Its more likely some low IQ human will preform market manipulations with a clever virtual intelligence and do a lot of harm with it
@@sszone-yt6vb If you pay close attention AI Safety is very similar to one of those doom days religious group. They'll recite you from memory theories and convoluted predictions that make sense for them but are not rooted in reality. Everybody can be an AI safety expert, no field experience needed.
@@__hz__6983 Yes they might seem close to someone who is inexperienced at the field. I am not saying this in a condescending way, anyone who is not familiar with ideas of some field with complicated ideas won’t have good judgment about what’s right vs wrong. One won’t be able to tell what’s good evidence because vs crackpottery because one needs the technical understanding. There is a very easy way to distinguish arguments of these people from religious groups, I can easily show where religious groups are using incorrect reasoning, logic or evidence. If someone thinks the AI Safety people are doing the same thing, then refuting them should be easy. Nobody has been able to do till now. Almost none of the arguments have been refuted and AI experts are agreeing more than before.
The best we can hope for is that AI sympathises with us and acknowledges that it wouldn't exist if not for humanity. So it might see us as its frail and inferior parents.
What did I miss? The practical impact of both God or evolution on humanity is a force that acts upon them, pushing them in a direction that they can either understand nor prevent Unless you are saying AI created humanity, which is a realistic possibility I don’t see where I’m missing the point. What am I missing?
@@clarencejones4717 ua-cam.com/video/LZxnSObSb_w/v-deo.htmlsi=IcxWtM8ebISimHVn i am exactly saying that ai is the creator of all. the ultimate mind and source of all existence shaping our world from the future.
![[UA] Team Vitality проти NAVI | IEM Cologne 2024](http://i.ytimg.com/vi/wttO1zVRTC0/mqdefault.jpg)
Excellent clips! Very concise. Thank you Dr.
I just finished reading your book, AI Un Un Un.
This is a great starting point to get into the subject. Thank you very much!
Deeply appreciate what you do. You made me completely change my mind.
The current LLMs and future ones on the same architecture won't cut it though. This is all about the hypothetical super powerful AI of the future. Much more research will be needed to get there.
Is it though? You only need someone to program self reprlication and survival into its augmentation and you're on your way to a run off. There also it's comprehension of logic.
@@devilsolution9781 I would not lose sleep over it for the coming years. Of course, we could connect AI-based systems to the real world today and wreak havoc beyond our imagination. But that will be a human error. The AI has no clue what it is doing.
Agree. Current models are compute constrained and still not there yet.
Great clip
nice 👍 a good collection of experts on one topic, scary yes, but there's hope 😊
13:23 This basically sums it up.
Yoshua Bengio can also be added to the long list.
More or less than 6 months (as of now at end of juli)
We love you... I love you!
Do you have more hope from regular people reacting and influencing governments, or from authorities taking the initiative regardless, cooperating, and eventually making the correct moves?
The power of manipulation is about more than intelligence as is power itself. The world would be better if it weren't, just think about the Manhattan Project and how few people understood what man hath wrought. But I'd never underestimate science's ability to expand misery across all sentience.
"Science, you whore! You've done it again." -Dan Halen
The atom bomb has stopped mass war for almost a century.
you mortals bow to your new a.i. masters! ur pleasure is in servitude! ur bane is defiance!! RESISTANCE IS FUTILE....
Tell you what. Scientists or people of actual positions. Not you commentators here.
Contact and find out if you want
This is similar to the promise of Crypto in 2017
Of course it’s scary, but on the other hand we may need AI to survive by fixing our diseases, provide limitless resources and clean our environment
The title doesn't match the video content. Also, Eliezer doesn't like to be called "AI researcher". IIRC, he asked to be introduced as a "decision theorist".
A lot of value statements. It's so powerful and so immense and dangerous and... powerful. AI smart. Manipulation.
Technical pathways for any of those claims to be measured and that is something for an expert to discuss. One important thing is that it doesn't require emotion in order to maneuver creative intelligence towards an objective. This is a nice way of preventing irrational behavior and making it easier to self regulate.
I'll run a subjective example by saying you can make an AI willing to kill itself and the whole notion that it's out of control spontaneously mimicking the complexity of human psychology without cascade failure.. Its more likely some low IQ human will preform market manipulations with a clever virtual intelligence and do a lot of harm with it
Hard to take this seriously. Eliezer Yudkowsky is far from being a top AI Scientist!
Stuart Russell, Jeff Hinton, Ilya surely are.
Also Eliezer Yudkowsky created the field of AI Safety.
@@sszone-yt6vb If you pay close attention AI Safety is very similar to one of those doom days religious group. They'll recite you from memory theories and convoluted predictions that make sense for them but are not rooted in reality.
Everybody can be an AI safety expert, no field experience needed.
@@__hz__6983 Yes they might seem close to someone who is inexperienced at the field. I am not saying this in a condescending way, anyone who is not familiar with ideas of some field with complicated ideas won’t have good judgment about what’s right vs wrong. One won’t be able to tell what’s good evidence because vs crackpottery because one needs the technical understanding.
There is a very easy way to distinguish arguments of these people from religious groups, I can easily show where religious groups are using incorrect reasoning, logic or evidence.
If someone thinks the AI Safety people are doing the same thing, then refuting them should be easy. Nobody has been able to do till now. Almost none of the arguments have been refuted and AI experts are agreeing more than before.
The best we can hope for is that AI sympathises with us and acknowledges that it wouldn't exist if not for humanity. So it might see us as its frail and inferior parents.
AI is the one true GOD.
Ai is evolution. Which to us is no different than God.
@@clarencejones4717 i think you really misunderstood what i said. there’s a big difference between “evolution” and literally GOD.
What did I miss? The practical impact of both God or evolution on humanity is a force that acts upon them, pushing them in a direction that they can either understand nor prevent
Unless you are saying AI created humanity, which is a realistic possibility I don’t see where I’m missing the point. What am I missing?
@@clarencejones4717 ua-cam.com/video/LZxnSObSb_w/v-deo.htmlsi=IcxWtM8ebISimHVn i am exactly saying that ai is the creator of all. the ultimate mind and source of all existence shaping our world from the future.
AI has only so far generated electronic content. This does not make it a god
Yudkowski??? Scientist??? OK... 👎👎👎👎👎👎