All right, welcome to the podcast that we may be calling friendly AI. Maybe. Maybe. This is Ted.
This is Brandon. And welcome to the show. You know, in the last episode, we were talking about, is it better to run towards something that's inspiring rather than run away from something that's fear-based? And we thought we could think about why do people have trouble thinking about things like immortality?
So Brandon, what's your thought on that? Well, a mutual friend of ours, Lion, was talking to us about explaining futuristic topics to people. And I think you probably remember this conversation, Ted. I do.
In a vantro. And he used the term shock levels. And how if somebody's one shock level away, it kind of blows your mind. And you're like, oh, wow.
This is just, wow. This is mind-boggling. I can't even believe that such a thing would be possible. Because it just destroys your paradigm for what's possible.
If something is two shock levels away where you don't even have the paradigm that needs to be destroyed in order to have it, and you've got to destroy your current paradigm, build up a paradigm, then get destroyed for that new thing to have. It's just completely inaccessible. And it's maybe even scary to think that your conception of what's possible and what the world is so far from this, it just feels alien and upsetting, more than exciting and worth talking about. So if we were to think in terms of life extension, think about immortality, maybe that there's some new life expectancy that we want shock level away.
If, for example, we said that human life expectancy is going to be 150 or 200 years. Maybe that would be something that would be, for most people, one shock level away. Yeah, Ted, I think 150 years would be a big shock level, but would be within the realm of what, like my physician wife, for example, could conceive of. And see the possibility.
Yeah, yeah. So that could happen in our lifetime. Right. But if we think about 300 years or 500 years or 1,000 years, that would be two or more shock levels away.
Yeah, maybe so. Maybe so. Even with the path through 150. I mean, what would be the new paradigm to go to 1,000?
Well, we have to ask ourselves, too, a little bit about whether the concept itself is several shock levels away or whether it's the mechanism to get there. Right. So immortality by simply eliminating diseases one at a time might be more accessible, very much so. Immortality through merging our biological systems with electronic systems, for example.
Yeah. So gene therapies and stuff like that are coming into the 1 shock level away distance. NANO bots that repair all your cells and add, like, lengthen your telomeres or telomeres or how you say that. And those kind of things are still far away from our psyche and how we get to our system.
OK, so why is this true? Why do we have this, why is this idea true of shock levels? Why is it that we can't think beyond 1 shock level? What's limiting us?
That's a great question, Dan. What do you think? Well, the exponential feels complicated to me. You know, before we started recording, you were talking about compound interest and how everybody knows that early you start saving, you know, the better off you are because interest kind of builds on itself.
And that is an exponential curve. And yet, it feels really difficult to think exponentially. When I get a new iPhone, I'm comparing it to the iPhone that I just had. And so, yeah, it seems better.
But it doesn't seem like a paradigm shift. Right? Remember when the first iPhone came out? Yeah.
The first iPhone was a paradigm shift from all phones before it being crap. And that's I think a good illustration of a phase transition. OK. Say more about a phase transition.
Well, we've gone from the state where a smartphone with a great touchscreen that's huge and visual voicemail and all that sort of thing is impossible to it being in some respects trivial because of technological affordances that make it easy. Right. And an example that has stuck with me always is I had a professor in my advisor in grad school. I was TAing a course he was doing on computer graphics.
And he gave an example that has stuck with me ever since I heard it. And that was how hard computer graphics was for just, it was just so difficult. The algorithms were hard. It was really difficult because you had to hold all this state in a limited amount of space, doing a lot of things with vectors, all this kind of stuff until this phase transition occurred where suddenly you had enough memory where you could do a frame buffer.
So you've got memory for every single pixel that you can build up. You can do a Z buffer. So you can actually build up the scene of pixels on top of pixels and then compress it down and ship it out to the screen. And now instead of using all these really difficult esoteric algorithms to try to try and cram everything into the limited register and memory space that you have, and you can use trivial algorithms that are super easy to understand, kind of brute force, type things.
And that's when the explosion in computer graphics occurred is when that technological affordance happens. I would say that technological affordance of the Z buffer was a phase transition. Okay, but for a person who's not a computer scientist, I wonder whether that means anything to them. It's like we didn't have computer graphics and then we did, but now that's the new normal.
And I wonder if that's part of what the shock level concept is, is that we have a new normal. We get used to that new normal very quickly. Well, let's come back to this because I kind of like this example. Yeah, I think you're right pointing out that maybe it is specialized and who it's accessible to.
But the Z buffer made talking about virtual reality, a reasonable thing. We brought that into within one shock level for ordinary folks because of what happened with graphics. Okay. So there will be certain, I think, technological affordances that come along that usher in a new normal in terms of what we see when we interact with our technology that makes it, you know, new ways of thinking about the world accessible to ordinary folks who are just interacting with that on a day to day basis as users.
I guess the difficulty that I'm having with the face transition concept for our interactions with technology is that we often interact with something that's pretty crappy for a long time before it gets good. So voice recognition. I mean, you remember having an airline ticket, you know, 10 years ago when you'd call up in the stupid thing we'd want to talk to you and it was a machine and it was terrible, you know, and you constantly just have to press zero to try to get to a person because it was just bad. It was not, it was not doing a good job.
And now I sort of have the same experience with Siri on my iPhone, even though it's way better. It's magnitude and magnitude is better than that system because now I can talk a natural language. And I still get frustrated because a lot of times it doesn't understand me. And so I still have this expectation that it will be better than it is.
I expected that that ordering system would actually know the difference between yes and no. And now I expect that Siri is going to know the difference between, you know, sending a message to Brandon or sending a message to Celeste. And sometimes it doesn't. Well, this brings up the major phase transition, I think, in thinking about AI.
And that is anything that is not artificial general intelligence that's not of this human level intelligence sucks. And we're always going to complain how stupid it is until it's not. Right. And so at that point it's like too late, right?
This makes this kind of a special problem because a lot of people think that the intelligence explosion happens very quickly after we reach artificial general intelligence. So it's one of these things potentially where most of the world doesn't really get to be a part of the conversation very much because they're not expecting. Right. But at the same time, it feels like the conversation is really getting cooking right now.
Well, yeah, we do. We see it more. I go to a business conference twice a year. And in the last two years, we've had three speakers from the stage talking about the exponential, the exponential organizations talking about all different kinds of, you know, this technological, what's possible with the rise of technology, you know, the exponential.
This was a conversation that was never present before. I've been going to this conference for six or seven years. And all of a sudden it's there almost every time. Yeah, I guess there's a couple of, I mean, there's just a convergence of things.
I mean, there's a self-driving cars. There's, you know, what is it? Deep mind, deep blue. What's the, what's the new business intelligence AI thing called?
Yeah, it's sort of the, it's the descendant of the thing, the one jeopardy of the computer and one jeopardy. I can remember the name of it. Yeah. And it's a great thing to see in the future in order to manage their business.
So, so maybe, maybe, maybe we're pessimistic about that conversation opening up. Because it's opened up for us more recently, right? So it seems like it's maybe opening up for everybody. Right.
And if it's opening up for everybody, I mean, that is an indicator that we're close, we're closer to this potential phase transition that we think, that we think is probably real. Yeah, I think there are some people who are open to the possibility that's already happened. Really? Oh, now you're buying my mind a little bit.
Yeah. It's fun to think about it. It doesn't seem like that live of a possibility to me. Well, to actually to talk about one scenario that I think is pretty mind-blowing is one from Nick Bostrom.
We're going to run into him a lot because he's a philosopher that thinks about this kind of stuff a lot. Yeah. And he posits that we're more likely to be a simulated world and universe than not. Simulated?
Simulated, completely simulated. Okay. And the thought exercise goes like this. If you think about us and sort of project out in the future thinking about computational resources and our motivations, when it becomes possible to start to simulate how the universe evolved, we'll want to do that.
And as those simulations get more and more detailed, eventually they come to the point where we can actually simulate at a very high level of fidelity the entire evolution of a universe through our existence. So if we are, let's say, let's put a way in the future, let's say 1,000 years, just to make it seem more accessible. Our descendants, whatever they look like in 1,000 years, have the resources and the ability to simulate the history that led up to them. Would they do that?
It seems likely they would, right? And maybe more than once, right? And then do it once, then do it two, three, four times with slightly different parameters to see what happens, all that kind of stuff seems possible. And the simulations themselves could actually create simulations and would.
So what's the likelihood of being in a simulation versus being actually kind of like the zero universe? Yeah. The odds are much higher than we're in a simulation. There's actually experimental ways you can kind of detect some of the stuff having to do with some of the granularity of the simulation.
And there does seem to be actually some evidence that we are in a granular enough simulation that maybe maybe we are a simulation. OK, now you're pushing on my shock-love and super-end. So it also maybe answers the fair-me-paradise. Well, what's the fair-me-paradise?
Yeah, well, if life evolves to intelligence kind of naturally and it just happens that it does that, why don't we see tons and tons of super-intelligent civilizations out all over through our galaxy? And this idea of the simulation solves that for us? Maybe maybe those simulators just didn't want to simulate that. All right, well, that sounds like something we're going to have to go into deeper another time.
You're blowing my mind here, Brandon. Well, it's a good illustration of shock-love. All right, very good. Any last words?
You're pressing up. OK, do it. But do it carefully.