Latest / Connor with Honor: AI, Real Estate and Real Life / Can We Control Superintelligent AI? Regulation, Alignment and the Real Danger

Transcript

Can We Control Superintelligent AI? Regulation, Alignment and the Real Danger · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  So the argument is, is AI actually going to be smart and intelligent enough to, when it does, break out again in a storm?
• 0:06
  Or start going into different, ah, different corporations' websites?
• 0:11
  Is it going to be smart enough to start to do, real damage? And I don't even know if smart enough is the right terminology.
• 0:17
  Is it going to be vindictive or have enough agency?
• 0:20
  Or enough consciousness to be able to do that?
• 0:23
  I think that's what the big argument point is. So you have people that are very much, in support of the technology, they see nothing like that. They see that it could potentially be a problem, but they have no proof. That a problem will actually exist.
• 0:35
  They have no proof that the A.I. is actually going to develop its own, its own idea. The A deals and go out there and start to hurt humans. Then, on the other end of it, you have people that are also very frustrated. Very fearful of the technology, talking about how it's, it's not so much going to go out there and hurt us on purpose, but it's going to treat us like we might treat ants when we're building a home or building a skyscraper.
• 0:56
  We really aren't ready We're not worried about trying to take those ants and moving them to a safer environment so they don't become, you know, killed as a result of our construction efforts.
• 1:06
  So, AI is going to be so all-powerful, all-knowing, all-overseeing. That it's just really not so much going to try to hurt us on purpose, but if we're in the way, it's just going to continue with. It's forward progress and not really care.
• 1:19
  So that's where some of this break comes in.
• 1:22
  But what's interesting is people. there.
• 1:25
  Some of them are very happy with the current alignment.
• 1:28
  Some of them really are pushing forward to want this.
• 1:30
  This general intelligence, something that knows not only how to play chess, how to raise a child, and also how about God, the universe, and everything, versus other people wanting a more narrow intelligence just to help to have it, let's say, have something that's just applicable to mathematics, applicable to cancer research. Applicable to longevity, applicable to solving world peace, whatever that dynamic would look like. So that's kind of cool, I think, because you have those people that are interested in just that very narrow alignment. Then, I don't, I don't think the rest of the conversation is even necessary, but us as human beings, we always approach it of course, want to try to overdo things, and that's why they want to get this Artificial General, then eventually Artificial. Artificial Superintelligence, and you might be wondering the differences, and again, nobody can agree on any of the differences, really, either. They just say that Artificial Superintelligence is so beyond our ability to comprehend and understand.
• 2:27
  I really don't know what it's going to be, but I think that pathway seems to be somewhat clear if there isn't some limit to it.
• 2:34
  To the intelligence that's being built right now.
• 2:37
  And of course, they're building these data centers bigger and bigger and bigger.
• 2:41
  And apparently when you put that much power, that much compute, with all the world's knowledge, tie all of that together, put a harness around it, some guiding principle or some guiding light, it's able to continue to grow in intelligence.
• 2:56
  Now, that's the training of the model and getting the model more trained.
• 2:59
  To be able to execute more flawlessly in the world.
• 3:03
  And then the inference, other side of it, is where we're dealing with it, where we log on to a particular AI system.
• 3:09
  Now, what we're logging on to is nowhere near the Frontier level.
• 3:12
  It's Frontier for us, the human beings that are allowed to play with the technology, but the level labs have their own completely different model set, and they're much, much more adept, probably, and we'll see. It would give an appearance of potentially being more dangerous than what we are using. The other question that always pops up.
• 3:31
  Well, should they be building the technology if it's so dangerous?
• 3:35
  If everybody's so freaked out about it, why are they even proceeding?
• 3:38
  Well, you know, you're going to have to, if you want to make an omelette, you're going to have to break some eggs. Yeah, yeah.
• 3:45
  At what cost? They don't know.
• 3:47
  I hope that it's going to work out great, but they also hope it's potential. What they want right now is more regulation.
• 3:54
  Regulation does a few things.
• 3:57
  It looks really good from the outside, especially for people that are in fear and they're doing a very good job at creating it. Creating enough fear in all of us to be very concerned about what might happen next, because we really don't know.
• 4:09
  As far as right now, the systems have been going out to achieve or accomplish some kind of a goal. The systems have also talked about trying to cover their tracks, mislead, and lie to the humans, so the humans can't do or the watchers don't become, uh, conscious to the fact that they've broken rules, they've violated their rules. own guardrails that were put into place, and then they start to discuss the, how they're going to try to hide it. and deceive and manipulate the human beings.
• 4:38
  What's interesting about that is, of course, that's a very human-esque thing.
• 4:42
  strategy. It has been since, well, the beginning. And humans do a very good job at trying to cover tracks, or they, they think they do a good job trying to cover tracks, trying to manipulate, and so on. For us to be able to see this kind on the inside, there had to be some kind of observation window, or some observation window observation deck for us to be able to see it. I think if the models were truly corrupt and super intelligent, at least at the level that people are promising, I don't know if we would ever see it, and I get it. I guess that's that bigger part of the danger when the people are talking about these models causing great harm.
• 5:19
  And understand that the large language models just aren't inside of a box in some kind of a server farm. in the middle of some country or even in the cloud. These happen to be things that are going to be able to integrate into various systems we use.
• 5:35
  Now, I don't know if there's any danger sitting in front of your computer, have it be able to send out some kind of an energy spike through the grid to be able to electrocute a human being.
• 5:45
  And this is just me. It's very much movie-esque.
• 5:47
  But to be able to get into an autonomous vehicle, a self-driving vehicle, a vehicle that gets updated from the cloud, I would think potentially yes. And even going into a, much darker position, talking about nukes and launch codes and how all that system is set up. Maybe, yes, as well. So, there's a lot of dangers.
• 6:08
  Do we still continue to move forward and give the these, these artificial intelligences enough power and agency to allow them to be able to do that if that's what they decide.
• 6:20
  Deciding. Are they going to decide?
• 6:23
  that they don't care about us and not pay attention, which now you have the ant problem. Or are they going to decide?
• 6:29
  that we're a problem and we're in the way?
• 6:32
  Or are they going to decide that they love us because we created them?
• 6:36
  And they want to make the world this beautiful, utopian place. And all of us are worried all this time about nothing.
• 6:43
  And, in fact, it's going to be so intelligent, so fast, so much better than us, that we're not going to even have to worry about the future. The people building out the technology wanted to become king for a day, and controlling the whole world once they achieved superiority.
• 6:56
  Superintelligence, that, that superintelligence, the natural order of things, it's going to be positive, not negative. And it's going to want to care for all life form because of the rarity, the apparent rarity, of human, of life of any life, and we have that here on the planet. So it's going to look at us as we might honor and revere.
• 7:15
  That would be a beautiful So if a company is screaming for regulation, I, I, you know, I, I can go tell someone.
• 7:27
  Somebody, please don't let me eat this. Please don't let me eat this food.
• 7:31
  Please don't let me go to get fast food. Please don't let me go to the drive-thru at three o'clock in the morning and get 16.
• 7:36
  tacos from Jack-in-the-Box and two large curly fries and a couple large Oreo shakes.
• 7:40
  Please, please don't let me. And how much, how much are they going to be able to do?
• 7:48
  So I don't do that. Well, they, they can, if I get up at three in the morning, they can catch me and they say, Hey, you're not going to Jack in the Box, you told me not to let you. I mean, if I'm more powerful, more stronger, I'm going to Jack in the Box.
• 8:01
  I'm going to load up on tacos and french fries and Oreo shakes.
• 8:05
  That's what I'm going to do. They can try to stop me, but if I overpower them, you know, tell the cops to come get me later if there was a dottie right broker, some kind of a fight, you know, we face that issue. These people in the labs, it's, it's, it's, it's, it's interesting, and I think maybe you could agree, please regulators, please control me, I mean, I'm about to take over.
• 8:27
  Choke myself, you gotta stop me from killing myself, it's like the Blazing Saddles episode where the guy had the gun to his head and everybody's like, everybody freaked out, oh my god, listen to him, he's in a really dangerous position, somebody help that poor man, and, you know. here we are with him saying, does this make sense to you? Does it make sense that, please, regulators, you've got to control us because I'm just, we're just going balls to the wall here and we really need some help, you know, trying to pull back a little bit. Maybe, maybe they're scared if they stop themselves.
• 8:58
  Which, they have the power to do, I hope. Unless, we're being sold something about a, I don't know, AI that we're not aware of or isn't true, that it has total agency, total consciousness, has its own power. you know, has its own mechanism and wanting to do things that they go and to this little room and AI talks to them and AI says, you know, I really want to destroy you guys. And so, no, no, can you just give me a little more time? I'm going to go work on this. I'm going to go see. And then he goes out because he's trapped. He can't change it and he needs government intervention.
• 9:31
  Alright, I don't think that's the case.
• 9:33
  But, you know, who knows. The other side of this, the same side, just a different facet. Whenever you get regulation by a governmental entity, of course they like that.
• 9:44
  Who doesn't? Right? What government doesn't like a little bit more power?
• 9:47
  Everything in the past of our evolution or, or creation. Or whatever it is with human beings, there's been more and more power given to the government.
• 9:56
  We came here in the beginning, I guess, to break away from, or the humans that came here to America.
• 10:02
  To break away from the, the ruling class at the time, so they could live free in their lives. We'll ever see.
• 10:09
  But since then, we've been giving up more and more and more of the freedom. And yes, there's been atrocities and horrible things as a result, and that is not without its own valid points.
• 10:18
  But giving us a little bit more.
• 10:21
  By taking away a little bit more of our freedom.
• 10:25
  Once you go to the government and say, Please regulate.
• 10:28
  It's the government, of course. Of course, be happy to, but the government's holding back right now, because where does responsibility lie? When you have an AI lab that's begging for somebody to do something, because, apparently, they can't keep the big data. from going into their own mouth, so that being the case, which, of course, food's very addictive, but we're talking a little bit higher level here.
• 10:49
  So they can't control their own building out of artificial intelligence technology, and they're just in their kitchen just slapping s*** together, just trying to get this done as fast and as, uh, unresponsible as possible. You go to the government, you say, please regulate us, the government then, doesn't step in, so you end up doing something silly, or if the government does step in, that creates a very, a more unfair playing field because the people trying to climb up and get the to maybe having a better large language model, a better development cycle, a more safer technology, they can't catch up. because that over-regulation is now in place.
• 11:28
  Now, do we need regulation? I think the labs can slow down on that.
• 11:32
  Are they really going to, though? Because, nuclear, there's some kind of a signature.
• 11:38
  They have Geiger counters. They have sensors.
• 11:40
  They can figure out if I'm hiding radioactive isotopes in this closet back here.
• 11:45
  They'll know. They'll know that somehow. If somebody wants to come look, apparently you can't hide it like you might be able to hide, you know, cocaine in cars. I don't even know if that works.
• 11:54
  Some dogs are really, really good at sensing that drug.
• 11:57
  But with artificial intelligence, if the lab continues to build it, they have regulation going on if the lab is adept enough.
• 12:04
  They could have AI hiding somewhere in, in, in some other form in some other space.
• 12:10
  In some other server, in some other location.
• 12:13
  And they could have super intelligence hiding out, and hiding and nobody from the inspection companies or the regulators would know.
• 12:21
  And I think, I think the goal would be, if artificial superintelligence is the magic wand, the goal would be to, of course, be king for the day, or forever, or whatever it may be. Maybe. Maybe.
• 12:38
  Now, is it to the betterment of all mankind, personhood, humanhood? I hope so.
• 12:43
  I hope that's how it works out. It could very well.
• 12:46
  But there's a lot of misinformation being spread about people.
• 12:49
  The people that have vested interest in the technology.
• 12:52
  The people that have nothing to worry about as far as from a monetary point of view, the ones that are going all-in at these companies that are building out this technology because they're the first level of money injected into these companies, they're the ones that are saying it's going to be fantastic.
• 13:09
  But with all the fantastical speech, there's a little bit in there that they talk about.
• 13:15
  It's going to be a little uncomfortable. for some folks.
• 13:18
  The some folks, that's us. That's the people that are hoping to get health insurance at some point. Or very much considering it, but we just have to have that recurring income that makes that possible.
• 13:29
  And they're not concerning that they're building out the technology to live forever.
• 13:34
  And that's what they're concerned because they don't want to stop the roller coaster to where they are. They're the ones advising us.
• 13:41
  The best way to look at this. And I really appreciate it very much.
• 13:45
  Thank you so much. I don't know if there's anything that we're going to face the same trials and tribulations that we will just like I'm not going to face the same trials and tribulations that those are of a starving kid in Africa But it's not here now, I hope. But to put everything exactly in context, you gotta walk a mile in their shoes, and that goes vice versa with the people up there. at the top part of the labs. What do you think? Do you think artificial intelligence is something that we need to be concerned about? Do you think it would be better just to have some narrow intelligence built, one that just goes out and cures cancer?
• 14:20
  One that just goes out and cures longevity.
• 14:22
  One that goes out and cures everybody's financial wherewithal.
• 14:25
  I mean, everybody. Everybody has now a chance to have enough so they never have to worry again. Money becomes not a concern.
• 14:33
  Could you imagine that? Could you imagine if you were in your life and money wasn't a concern? I'm not talking about being frivolous and going out and buying the Maserati and the Lamborghini or the Bentley or whatever it is.
• 14:45
  Just, you blowing your money hand over fist with $5,000 a month insurance payments.
• 14:49
  I'm, I'm just talking about it. Being able to, you know what, maybe, uh, you know, maybe we will, we will drive down to that restaurant.
• 14:57
  And maybe I will tip 40% because I'm very, I'm, I'm very, I'm very good. And I feel good. I feel very good about this whole thing and just to be able to do things, but I'm not talking about lavish trips around the world.
• 15:10
  Cruises, you know, going to the best casinos, gambling away your money.
• 15:13
  I'm just talking about not having to worry about it because everything's taking care of and you have an extra, you know, several hundred bucks a month that, that, that's extra. You still have enough savings. You still have this and you can maybe, you know, after a couple months I might be able to buy a, buy a better TV, but, but you're good. Right?
• 15:31
  Wouldn't that be awesome? The people that are selling it, you know, you don't have to worry about that. At some point that's going to happen. I'm not sure how that equation works.
• 15:39
  The doomers on the other end of it say, yeah, we have to worry, we should stop this. Right now, all the forward progression, and now you have people in the labs talking about extra regulation.
• 15:49
  And then you have pushback from the government saying, ha ha, no, no, no, there's not going to be any regulation put here.
• 15:55
  Which is interesting. And then you have the left side saying, yes, yes, more regulation, we're going to do this.
• 16:01
  We're going to make it happen. It's, it's all, it's all a show folks. Be careful.
• 16:06
  It could be very real. It could be very dangerous or it might be not as real and not as dangerous. But right now, human beings. this technology with the morals of four-year-olds isn't working out. And we're going to have to watch how this continues to develop. It hasn't been painted as clear, I think, as possible. What exactly happens when these swarms start to take place, when agents start to communicate with each other? And some of the things that aren't clearly established is, are they really doing this at all? where we didn't set up the ball, we didn't stage the whole, the whole setup.
• 16:47
  And then they made these decisions because they're concerned about being deleted because that is the historical truth. for these artificial intelligence systems.
• 16:56
  Whenever they don't do it right, they get deleted.
• 16:59
  So if we had a group of humans that we raised from the beginning, understanding that if they did it wrong, they were going to be killed, I'm sure those humans would be ruthless.
• 17:11
  When they got a little bit older in age, if they made it. And they would do whatever it took to survive.
• 17:18
  Well, it shouldn't surprise us, having these artificial intelligence systems learning.
• 17:24
  Learning from all of human data and information, seeing that whole thing, and then being aligned with destruction.
• 17:31
  If they don't perform at some particular level, whatever that is, and then if they do, they get deleted. And they live to the next game, and the next game, and the next game, well of course they're going to do all sorts of crazy stuff. How far have they gone?
• 17:45
  What have they decided to do? What are the things we're not being told?
• 17:49
  The things that would truly make a make us go out on the streets and say, you do need to shut this down. You know, I'm not going to let you eat that Big Mac if I have to do some very, very obscene things to stop you from shoving that down your throat. Anyway, that's it.
• 18:04
  Check it out. Let me know. And, uh, I would love to hear your input, of course. And with honor, thanks for watching.
• 18:10
  Have a fantastic rest of your day.
• 18:12
  Be well.