Welcome to concerning AI. I got to start on my whole mind. I can see you. Exactly.
We're just kidding. We haven't started yet. This is not concerning AI. This is leathery light.
Woo! It's spreading now. Nice. Now I can see you.
See you. Just ignore that first part. We didn't mean it. Because I'm an embodied AI.
Or an embodied intelligence. I have to have my stimulus. Nice. Yeah.
Welcome to concerning AI. This is tight. And this is Brandon. Yeah.
Feels like I'm taking a little pause because it's been a while since we've sat next to each other. You listeners might not know that because our recording schedule is not necessarily connected to our release schedule. But we are embodied and we do sit together on the capture time. It's fun.
I like to sit with you. It's really fun. It's been a while since we sat together. But last episode and this episode might come out in the same week for all we know because editing happens at another time.
That would be amazing. That'd be fun. Last episode, we began talking about what our current kind of narrow way eyes and let's project into the future one of those and just see what might happen. And then it seems like we have maybe some other episodes to talk about.
Last episode we talked about tools and what might happen is they get better. Could they become a path to AGI? Or to super intelligence? And with tools we thought maybe not.
Okay, I'm glad you're reminded me here. So I appreciate that. So what's the next current kind of narrow eye that I think we should just skip all of the other stuff for a little while and talk about robots. So not robots, but embodied intelligence.
Yeah, so I think robots are really fascinating because we do have programming of devices that have some smarts in them now that we call robots. Things like industrial robots are not actually robots as far as I'm concerned. But things like Roomba is an actual robot. You sort of tell it, go clean the floor and it does its own thing.
It figures out its own way. Yeah, it has a really great sense of humor. Roomba does. Makes you laugh.
Yeah, I think humor is actually, it might be a good. Because humor tends to be so absurdist. You kind of need metaphor and then, well, I don't know that you necessarily need metaphor. But some of my favorite humor is metaphorical.
It's like taking some situation that we all this really get and then adding some absurd element to it. So as you think about current robotic technology, how are we doing on humor? You know, that's a good question. I'm not sure that any robots are intentionally funny.
Yeah, but I know. But that doesn't mean that that just may not be what I gravitate to. Maybe I gravitate towards the serious. Okay, so do you think that humor in robotic artificial intelligence is going to come spontaneously?
Or is it going to come from some programmer thinking it's a good idea and putting it in their own purpose? Do you think it can emerge? Maybe. But let's talk about my youngest son's first joke.
What's this for sure? Because I feel like it kind of illustrates how humor can be embodied. So my youngest son's name is Ren. And when he was a baby, couldn't talk, couldn't walk yet.
He made his first joke. And the joke was he was out in the backyard. We were out in the backyard. And he was kind of crawling around and there's grass and there's rocks and there's stuff.
And he picks up a rock and he starts to bring it towards his mouth. Watch it, Shelley. She's like, ahh! And then he laughs.
He actually puts it in his mouth. Then he laughs. And then he takes away. And then he watches her.
And he does it again. And he laughs. And he's watching her and getting her reaction. And he's doing something.
He's having the communication with his environment and with another person that's also situated in that environment. It's all physical comedy. There's no words happening, but meaning is happening. Well, Coyote and Roadrunner.
It's completely physical. There's literally no words of B.B. That's a sound though. It's not really a word.
Unless they hold up the sign of this beat, which is funny too. Right, right. I mean that's physical humor. My kids are about to turn 9 in two days.
And one of them absolutely loves physical humor. And he laughs and laughs and laughs at anything that's physical in a movie or a TV show that has that quality to it. He likes words too, but he likes the physical stuff. Yeah, I feel like physical humor is kind of like an early form of humor.
Yeah, you wouldn't say that the three stooges were high art intellectually. No, it's not literature. It's not like tickling our frontal lobes. But it's connecting with that kind of inner brain.
And it's funny. So do you think our Boston Dynamics robots can develop humor? Yeah, it's funny that you asked that. You've seen that the parodies of the guy pushing the robot down and like what he says.
Yeah. So somebody watching that can certainly develop humor. Right. Yeah.
That's a good question. Well, I mean, it kind of begs the question a little bit about what meanings are the lowest level, the most primitive. Because it feels to me like when I look at Boston Dynamics and these other robotic systems, it feels to me like a lot of embodiment has been skipped. Like they've just jumped straight to trying to solve some tasks without, you know, if you look at the evolutionary pathway that leads to humans in body, it doesn't start with big old limbs that you can move around that can't feel anything.
That's not where we start. You know, there's all of this touch and proprioception and I guess proprioception is actually something that makes sense. I think there's been a lot of work done in the area of touch. And I don't know what touch means to the machine.
It's hard to tell. But there's been a lot of work done picking up things that are very fragile, for example, where the robot has to have a sense of real gentleness to its grasp. Right. We're an industrial machine.
You can grasp it as hard as you want. If you're trying to paint the car, you grasp the paint nozzle as hard as you want. Or you build a nozzle right into the hand, right? And it's not even grasping.
But if you want to build a machine, you know what, if you're going to pick up the car door and put it on the car, it doesn't really matter how hard you pick it up. But if you're trying to pick up a light bulb, or I don't know what else they've been working on, but they have been working on that stuff. Yeah. Don't you think that being gentle is being appropriately firm?
I don't know what words they use for that scientifically. Yeah. That's exactly in the realm of, I mean, you think about the meaning of squishing something. Like there's a lot of rich analogy that I feel squished when somebody says something mean to us.
Right. But you know, there's a lot of in intelligence that comes around being gentle. Right. And you know, so much of our emotional interactions with each other are characterized by petting and that sense there.
I mean, you know, like gentle touch. Right. And they were gentle with me or they were rough. You know, so I do think, I do feel like there is a lot that is rich.
That feels rich. Yeah. We talked about reinforcement learning last time. And I wonder if they're training these arms to pick these hands, to pick things up.
And basically try to figure out what's the appropriate level of touch for different things. Yeah. See, I kind of feel like these are a lot more still, a lot more one off, sort of hard program, you know, programmed more than learned more than less embodied. I just don't have the sense that the robot hand really has a sense of self in its own presence and how soft things, you know, it's kind of a gimmick a little bit to pick up the light bulb and screw it in.
It's kind of like somebody's done some clever programming to just barely eat that one task out. I feel that way. I don't think they're programming the hand to be the light bulb screw. I think they're programming the hand to learn how to pick things up and to fail and to get better.
Yeah. I think I'm a little more cynical about the progress that we've made there than you are. I feel like there, you know how people sometimes talk about Facebook, sort of creating the sense of everybody's lives being so much better than they actually are. They actually showed a good thing.
Yeah. I kind of feel that way about this embodied stuff too, like the light bulbs and stuff is that they, you know. Okay, but if you take a breakthrough that we've had with AlphaGo, for example, and you take that and you put it into your embodied systems, into, you know, could you start with a hand that isn't programmed at how to pick up anything? Like AlphaGo wasn't programmed on anything other than the rules of Go.
It wasn't programmed on what we as humans think of as good Go moves. And so what if we don't tell it in advance? What heavy thing is or a light thing or a soft thing or a fragile thing or a what if we don't tell it that, but we let it, we give it the same kind of reinforcement learning qualities and we let it experiment. Can we get there?
Let's say we're not there today. Let's take it as a given what you say. Yeah. We're not there today.
But if we use that same technique, can we make advancements with embodied AI that way? Yeah, maybe. I mean, again, I feel like we're jumping over all of these intermediate layers of meaning that are inside of picking up something like that people do or any animal that picks things up. There's just this huge journey of evolution that started with something so far removed from that and so much more primitive.
I worry that we're jumping that we're jumping past the, that we're missing that actual sense of touch and stuff that we're just jumping straight to performing a task without any meaning sort of layered into. I'm trouble articulating this point. Push on me here. Well, you're like, I feel like you're assuming that that meaning has to come from developing in a certain way.
And I guess I'm just trying to speculate on what we have today and what we might be building next and where we go from there. Like, can we get, I've got, okay, so I've got a room, right? Which is for those of you that don't know, it's a very simple robot. You put it on the floor, you push the button and it cleans the floor.
It just bounces around the room. It bounces around the room. It does map the room. It sucks and it bounces around the room.
It sucks and then when it finds a dirty spot, it knows that. It taps the room. You think it maps the room? I think it does.
I think it knows over time where the walls are. It tries to be efficient in how it, I don't think so. Okay, well, that'd be even more primitive than I thought that I thought it was. I think it knows where its base station is and that's about it.
But to know that, it's got to know where the rest of the room is. No, no, all it has to do with know is where the radio transmitter is coming from or some beacon. I think in order for it to know that it's covered the whole room, it has to map. It doesn't know that it's covered the whole room.
It knows when to go back to its base station. Well, it needs to charge. Yeah, but it won't just keep, it won't keep trying to clean the same spots after they're clean. I don't think so.
Oh yeah, it doesn't know. I don't think it knows. Maybe they've come a long way since I last looked into them. It's admittedly five years at least.
I think it does not re-clean areas that it's already cleaned. I think it knows. Maybe, maybe. I think it's pretty efficient.
I think it's better than you think. Better than you think. Yeah, I don't think so. We have the actual robots that you want to point to as useful to talk about it all.
Because so far everything I've brought up, you've pooed and said we haven't made any progress in the world of robotics. Wow, man. Yeah, that's like isn't improv you're supposed to say yes to something. I'm saying no to everything.
I'm looking for a place to start from so that we can speculate into the future and so far we don't have a place to start from. Well, all right, so we can go back to Rodney Brooks from 20 years ago. Okay. And he was involved with that.
No, he was doing these little bugs that were kind of like, okay, so I liked his approach. And so maybe I'm still infatuated with his approach that honestly how far did it really go? I don't know, probably not that far. Although Roomba comes out of his company.
So let's maybe I'm not so far. Yeah. So he made these little bugs. He's trying to make him body intelligent.
He's kind of one of the fathers of the idea of embodied intelligence in the first place. And so iRobot is his company in Roomba and I think a couple of other things. So the idea is to try to make something that can actually make its way around the world in a very, very primitive way without trying to make it quite so, not trying to make it accomplish high level tasks that we think of as high level tasks. So I like that.
So I like to think that the thing bumble around and find out what the world is from its bumbling around with whatever sensors it has. So that's an interesting way to develop the idea of embodied AI moving toward AGI, but that doesn't accomplish anything useful in the world, which I thought is what we were going to talk about. What are robots actually doing today and where are they going to go? I'm saying I think we're going to dead end until we actually pick this back up and follow that slope back up.
I think we get to all of the rich meanings in the world unless we actually have that we take for granted and that we love, unless we have something that kind of follows the same developmental footsteps that we did. It's not going to have the same meanings. I don't think we just give the meanings to something and have it have it have it have those like how you know what does it mean to to you know why we see anything that we look at almost we can say this is the front of it and this is the back of it. Where does that come from?
That comes from like seeing the front and backs of lots of things and experiencing that one right there. That one like where can you how can you get that meaning does that come from the fact that people have fronts and backs and that we're so good at our visual cortex? Animals do right? Not just people.
Well, so faces I mean things that faces it's really clear what the front isn't with the back is our facial recognition part of our brain. Is that where that comes from the front of the back as a as a result? So I want to say that it comes from some sort of empathy that we project ourselves into all kinds of things and that that's part of how we model the world. Like if I were that thing that would be my front that would be my back.
This would be close to me. This would be far from me. So we have this natural sort of built-in sense I think of like what is close to you know like maybe that toy car or something like that. You know like what's close to the toy car?
Well, can the toy car if I were the toy car could I reach out and grab it or would I have to move toward it to get it? So I feel that we break up the world into these interactions where we think about them, you know, empathetically and that's the main way that we reason about the world. Like if I'm that and then if I'm this other thing and and we're all all these other things that I'm pretending to be are interacting with each other. What does all that mean?
So like to get an embodied sense of near and far I think requires that you learn what you can reach and what you can't reach. What you have to move toward what you can't move toward. I feel like we're talking about us who already have a general intelligence and I'm not I'm not clear on the path between the before and the after. So I'm giving two steps here right.
So one step is to learn for me what it means for something to be close to me as some little thing right. So again, I empathize, you know, because of how I'm putting myself into the little insectoid robot and thinking about like it learning and stuff because that's the sort of intelligence that humans have. But where does the concept of near come from from a little bug robot that is just bumbling around the world? This concept of near is that something that needs to be programmed or does it emerge?
We have this concept of near but we're already a general intelligence. So the concept of near is just a symbol that we attach to something else that we've learned over time that has to do with other goals like eating things and stuff, right? So we have some goal that's programmed into us by natural selection to eat stuff and to avoid being eaten. So like our notion of near I think comes from having learned that over time.
I want to eat this. Is it near? I'm using language here. Forget language for a second.
Like I'm in the world and there's some things that look tasty or that seem tasty that smell tasty that whatever, you know, which one first? Which one do I go after? Which one is near? Like which one do I think I can grab?
So coming back to the little buggy robot kind of thing, it has to have some kind of goal programmed into it and then the ability to bumble around its world and accomplish that goal more and less effectively. Maybe. I don't know. I don't know Ted.
I can say no, but I can't say yes. That's terrible. So I have a very different sense of near far recently because I have crossed the United States twice once going from Oregon to Massachusetts and the other time going from Michigan to Oregon. And it used to be that I would think of a drive that's three hours as being kind of a long way.
And once we did that drive where you're tearing off 12 to 15 hours or 18 hours a day, then when we got to a place where it was like, oh, it's only four hours. Super near. You're near then. Yeah.
So I think it's really fascinating that it's completely context dependent. So I think that's supporting what you're saying in terms of when I'm a person on foot. I'm an ancient person right on foot and I eat the things I can get to on foot because I walk around. I'm not going to go every day to the place that's a three day walk from here.
I need to situate myself near the stream because I need to drink every day multiple times a day. I can't situate myself three days walk from the nearest water source. Right. Because that is a different thing.
Now, I might only go above a hunting a couple times a year, but they're far and I could go and there's different contexts for different things at different times. So maybe a way of saying near is at hand. I mean, that's a more basic version of near. Maybe I think there are people who are on airplanes a lot like every day.
And for them, if they live in New York, LA is near or even London is near because they're just used to getting out of plane and they know it takes a certain amount of time. And you deal with that. But for other people, people that don't have access to an airplane or even a car, the places that they can walk to in their neighborhood are near to them. It's a different sense.
So I wonder what some of the fundamental concepts would be other than near from being in body. Near is one that is near and near by car. I see people who are like, oh man, I love that. Yeah.
We saw some deer on our trip too, but that's a separate idea. Oh man, that is sweet. We also saw tens of thousands of prairie dogs. I had no idea there were so many prairie dogs.
What do you think near means to them? It was really fast. So you could kind of measure this because as you walk closer to them, they would do different things. So at some point, you see them standing up next to their whole and they would chatter to you.
I can't do an impression of a prairie dog, but they would chatter to you. You can tell they were reacting to your closeness to them. And then at a certain point, you got too close and they just disappeared down the hole. So there was two different sets of near.
One set of near was alarm. Sound the alarm, sound the alarm, sound the alarm. Let everybody know. There, there, there, there.
Hey, this guy, look out. But then when you go past a circle line, they were just gone as a safety measure. They were just under the ground. And they didn't hear anymore.
They got quiet. And then you can walk all the way up to the hole if you want to see it at that point. So they were fastening. They were super cute.
They're little. They're only like kind of squirrel sized. Smaller. Oh, really?
Well, no, it's a squirrel size maybe. I never saw a squirrel stand up on a time like that. So you can't... It's hard to judge.
It's hard to judge. But there are tens of thousands of them in these colonies in South Dakota. It's fascinating. So I don't know.
They're on their way to language right now. They're working their way towards it. Very social creatures. Yeah.
Maybe they have language. Maybe they were very connected. They seem to be very connected to one another. Yeah, there's this.
Look out. So what are the other concepts other than the primitive that might be building up? Oh, see now. This feels tasty.
Tasty. Yeah. Does it taste good or good? Yeah.
I like how that came out there. It feels tasty. And then maybe that's a primitive. Tasty, tasty.
Tasty. Things that are tasty are probably not poisonous in general, I would think. A lot of calories. Well, for...
My kids picked a fig off a tree walking around town today. Yeah. And it wasn't ripe. And they made a real bad face and spit it out right away.
Not tasty. Not tasty. Not tasty. Not working as well.
Yeah. Mmm. And so tasty feels like just, again, this is not like, this is kind of thinking based on what you need as an organism to live. What about safety?
Is that primitive or not primitive? It feels really basic. Is it made up of other things? Mmm.
Maybe like, I don't know, cozy, defensible. So, you know, like the whole space bubble thing. At risk, kind of thing. Like, they're going to, well, like, you know, there's things that get where they get too close or threatening.
Right. When they're far enough away, I can just notice them. But if they're too close, I got to do something. So, that's the nuance on near.
That is another danger coupled with distance is a thing. So, I think maybe agency is a thing. Like, that thing is moving purposefully. Mmm.
That thing, you know, like a tree can be near and we don't care. But not threatening. But the tree is not moving in a way that, you know, I feel like, maybe it's kind of more like faster than me. Like, I can't get away.
Again, it's kind of, it's all related to kind of near too close for comfort. It's kind of, it's another nuance. Yeah. Yeah.
We're talking about the nearness of a food source, but there's also the nearness of a threat, which is what we're seeing from the prairie dogs. Mmm. Different reactions based on. So, we have perceived distance.
Right. Okay. So, it would be an interesting experiment to not try it, not with a human, but try it with some other creature that was faster than a human. Would their distances be different?
Presumably they would. The prairie dogs would know to sound the alarm earlier with a creature that is faster than I was and to run into the hole faster than they did with me too. Right. Like a shadow going overhead would probably immediately create a different reaction than the, I'm going to yell about the shadow.
I'm just going to just, you know. That's right. A bird. Right.
And then I'm going to be gone. And they don't care what kind of bird it is. No. Shadow moving quickly, right?
Yeah. Well, this is discretion. The discretion is the better part of our. Yeah.
So, interesting. So, we still, have we, have we, what have we come up with here? Not much. We still have near, I think.
Near is pretty good. Near. Yeah. And near is related to these other things.
Safety and. Near seems like one of the most important things like it is because it's related to energy, you know, too. Like, and we don't have that, you know, we got to be really efficient. Efficiency seems important, but not as a, as a something that we perceive directly.
No. And there's more of the outcome that a lot of our perception has pointed at. Yeah. So, tasty is kind of like, this is efficient because there's a lot of energy in one, you know, like dense, densely concentrated energy here.
This is efficient. So, tasty and near both feel like energy maximizing primitives, right? And they probably, if we keep, if we keep using energy, magnetization, maybe that'll help us. But they probably also are danger minimizing at the same time as they're energy maximizing.
Yes. Danger minimizing. Yeah. And energy maximizing as well.
Like, like efficiency. Like you don't want to waste your energy because that's another way of dying. So, that's, is running out of energy starving, basically. Right.
Starving sucks. Getting eaten sucks quicker. Yeah. But no less or no more.
So, you know, you would definitely, you see what animals do. They'll put themselves in a lot more danger when they are getting down to. That's right. The more hungry you are, the more risky, the more risky actions you can take.
Yeah. Because the cost benefit analysis. Okay. So, hungry.
Yeah. Hungry feels pretty primitive. That's an interesting one too, because that's not one that I think of when I think of the body, but that's a pretty, that's a strong sensation that is built into, it is built into human animals for sure. I assume that other animals have that.
Yeah. It's fascinating. I mean, I would always assume that too, but now that we're talking about it, I wonder. They must.
They must feel hungry because they have something that's like that. They have an imperative to get food when they're running low on energy. Right. So, that's so, we'll call that hungry.
Maybe it feels different. Maybe, but seeing red, you know, what you see when you see red might be different than what I see. Don't, don't, don't even start with that. But hungry.
Yeah. I think hungry is good. I like hungry. Yeah.
Okay. Maybe pointed it. That's a weird one. I never would have thought of that one.
Hungry. That's a pretty sexy feeling sexy. Yeah. That seems like a primitive one as well.
Does it need to be for an AI and hungry? Well, hungry, like an AI needs to have energy. If you don't understand hungry, you don't understand sexy. There's a ton of language you're not going to get.
Well, right. But so hungry seems like it translates over a right way because you have to be with batteries or you have some energy source. Okay. Sexy feels like not obvious that it would be.
I see what you're saying. I see what you're saying. Yeah. Whereas near would be and safety would be, right?
If the AI has a goal and something that is making it harder to reach its goal. It's a safety doesn't, safety doesn't seem that important either because, you know, your, your, your person like replication and safety are care, are somewhat cared for by your maker by the person who's watching out for you. Well, if, okay, if the AI though, let's, if we build, if we build the prairie dog AI, okay, the prairie dog AI, it's goals. Let's say we program it in, you need to satisfy, you need to have energy, you need to get energy and you need to not be eating.
Let's say we, let's say we program it. Yeah. And then the prairie dog is going to figure it out. And so the prairie dog is going to have hunger, which means my energy is running low and you need to get more energy and it's going to have near and it's going to have evaluate the threats.
It's going to have all those things. Okay. Automatically. So if we did, if we did some sort of genetic algorithm for prairie dog like robots and there was a power source and there was a, what kind of predators would there be for the, we had to come up with something.
Yeah. Birds are not actually going to eat. We put them in an environment. I mean, for the purposes of the, of buying AI, they have to be in an environment, a real one or one we create in order for this to make any sense, right?
Yeah, right. So we put them in an environment that has predators, some kind of predators. So they have to, we want them to be like prairie dogs. We want them to learn like prairie dogs and have the embodied concepts that prairie dogs have.
And it feels dangerous to me because we'd have to create some robots that eat robotic prairie dogs and what they, those ones involved in eating us. It's all dangerous. This is why we have concerning AI because there's so many threats that we don't make. The little prairie dogs, I'm not too worried about them.
Let them rule the world, right? It's the things that eat the prairie dogs that I worry about. You're going to really meta on us. Yeah.
Well, all right. Now we're coming back, we're looking at it all back in here. But sexy is where we were trying to get to it. And it's actually doesn't seem automatic to me.
The replication, okay, doesn't seem, so we make up, okay, we have to make up this world. You know, and in this world there is, there are some rules. The way the thing works, right? Maybe if we try to actually, instead of simulating, if we try to actually do it in some sandbox somewhere or something, right?
Right. There's a spot where the robot can actually plug in to charge it's battery. Yep. But then we also have to set up some rules about like which ones get, like which programs, okay, so let's say they all have the same chassis to start with.
We won't have physical evolution to start with. But the programs are going to evolve. Yep. And there's going to be a fitness function sort of thing.
So which ones get to evolve? Well, clearly only the ones that don't run out of energy, run out of power, you know, like that, like if you run out of power, you're dead. So you're going to, at that point you'll have no more, no more opportunities to what, to present it in the down line. So somehow there has to be some kind of reproduction happening.
So either we're looking through and we're finding the fitness according to us and then we're doing some sort of asexual reproduction of those. Or there is sex and they find each other and do some kind of exchange of code or some kind of exchange of something, right? But this, I mean, this is way more complicated than the idea that they just get to run trial after trial for trial themselves to see which ones are best. Yeah, but you've got to pick the winners and then improve, you know, like you've got to have generations.
Well, I'll go to those now, generations. Oh, it does, absolutely. No doubt. There's like a whole cohort and then they take the best ones of those.
Really? No. Well, there may not, there may not be any sort of genetic. It would be cool if it did.
It would be cool if it did. But they're certainly, they're certainly taking the best ones and the best ones are playing each other. Yeah, no, it is a, it has to be, I don't think it's genetic algorithms in the, in the purest sense, but the best ones get to play each other and then tweak their parameters. So it is, it's, it's gotta be like the, the, somehow they're searching the space of all, of all parameters spaces for their functions.
Yeah, we've talked about that at all. We haven't talked about that, that part of Africa, of Africa is progression of how it gets better. We've talked about it playing itself, many, many, many, many, many, many games and getting better by having more experience in the world. But we haven't talked about culling code or, or even variation.
Yeah, I mean, I guess reinforcement learning is, is a kind of search algorithm that is, you know, so you have the same, the same basic one that is, is finding better and better parameters. So you've got a pool of these that are each one finding better and better parameters as they play each other. Yeah. And so, you know, it gets to the end of a game and it says, what I did this time was good.
I'm going to do more of that in the future. It's kind of what reinforcement learning is. So, you know, maybe, maybe that's not a genetic algorithm per se. Right.
Okay. So, I mean, for, for creatures, for humans, you know, for DNA based life, yeah, sexy has got to be one. No, no, no, no question about it. Right.
So, whether we engineer that in or create the space of it emerges or whatever. And can you have something that we would look at as actually like, as generally intelligent as we are, like who we communicate in a way that we felt like, yeah, that thing is that smart. If it didn't have that, well, like it seems like there's a lot of meanings and stuff that are important to us that I don't have. We are at the point where we're potentially creating immortality for humans.
We may, we may run out of this. We may go the other direction if we evolve ourselves into immortal spaces. Sexy and probably always still be there, but reproduction might not be interesting. Yes.
Sexy, sexy wouldn't necessarily need to stay important. Right. I've been thinking a lot about the stories that our world religions have always told about the importance of death. And I've been wondering, do we tell those stories because we don't know any way around death and, you know, the importance of death.
Yeah, death is absolutely critically important for the evolution of the soul in many of our major religions. It's absolutely essential that death happens because that frees the soul to be in its pure form. And so many like a manipulation, a classic manipulation of the system, you know, like the social systems would want people to feel better about how life is. Just some sort of parasitic effect.
Yeah. Where death anxiety is a, is one of the primitives. Yeah. A death anxiety being a primitive makes sense, right?
Absolutely. I mean, I don't know if that was the same as safety. Is it? Yeah, it's kind of the same thing.
It might be the same. Similar. I mean, yeah, I mean, maybe not. Maybe death anxiety is a uniquely human thing in that we are looking forward to death versus the immediate safety.
That's right. And safety, even the sense of future might be different for humans rather than anybody. Any other creatures might not have any sense of future at all anyway. So are we trying to create something that we would call artificially, generally intelligent?
Are we trying to call it create something that we would feel like could understand us fully? Because I think I got lost. I think I sort of ended up sliding into the second one accidentally. Yeah.
What I think is interesting is thinking about the kinds of narrow ais we have today and projecting into the future what could be the way that they get better and could artificial general intelligence emerge from any of them. So that's what we talked about tools last time. And it didn't seem like tools was a path by itself to AGI. Yeah.
Meaning is such a slippery concept. You know, I want to dismiss those things out of hand, but maybe they are creating a form of narrow meaning that it may not be accessible to me, but maybe it is a genuine and real meaning. And that as they as that sliver of meaning broadens. You're saying to what robots we talk about the tools.
You know, maybe that meaning is legitimate too. And that it can sort of broaden out from that sliver of meaning in the same way that maybe life broadened out from like a photo receptive freckle that could sense shadows into becoming the eye eventually and all the meaning associated with shadows like, ah, runaway and maybe that sliver that we have is not privileged. Like I'm trying to say it's privileged that you got to be embodied and sort of recapitulate the evolution of humanity in order to have general intelligence that we would see. But maybe that's not right.
I mean, I think plenty of people say that's not right. For somebody who's not hung up on it. Well, just to recap then where this conversation sort of came from and where I think we could take it. And then we talked about the tools are kind of narrow AI today.
And we talked about what's in an AlphaGo. Lots in an AlphaGo. That's the thing. And then search it.
Yeah. So robots maybe we got off track today. Maybe we come back to robots. Maybe not.
Maybe we're done. I don't know. I'm not sure if we're done with robots or not. But robots and where they might go.
Siri or Alexa or other personal assistants where might they go? Oh, man. Yeah. So I think that's an interesting conversation.
The games are kind of in that same route bucket probably to like non player characters. Yeah. Right. Right.
Non player characters would be interesting to see. Does it fit there? Does it go to a separate place? Yeah.
So we have today that we're going to see improve over time. Yeah. I don't know. I can't remember.
I thought I had six. But no. Did you get search engines? Search engines.
Yeah. I mean, okay. So let's put search engines in there. I don't know if that's just a tool.
That's a separate thing. Yeah. Okay. Maybe that's a tool.
Although what about like social networks or other like big data, you know. Things that operate on a completely inhuman time scale and scale and scale. We don't have any sort of intuitive connection with it's very hard to empathize with social network. That's right.
That doesn't mean there isn't maybe something there. Right. So we're going to talk about a book that we're in the process of reading and there's some stuff in there about that. Oh, and I think collective intelligence was one of the.
Collective intelligence. Absolutely. Simulation. Simulation.
Simulation. Simulation. It might be one of the most important. Yeah.
We'll also we'll listen to last episode. You all are laughing at us right now because you know, you've recently listened to last episode. You remember better than we did what we said. So we'll listen to it and we'll get back to you.
But we've got several things in the future and maybe after we do those or before we do those robots, I'll make more sense to talk about again and maybe we'll move on to something else. Yeah. Maybe we'll make some virtual skin or not virtual but some some real skin for feeling for feeling. Yeah.
It's important to me right now at this moment. Oh, yeah. Wouldn't that be fun? I like my skin.
I like to have it. Yeah. It's nice. Yeah.
Okay. Well, until next time if you want to get involved in this game, you can send us a voice memo. Skin on the game. Yeah.
Yeah. Yeah. Yeah. Yeah.
Yeah. Yeah. Yeah. Yeah.
Yeah. Yeah. Yeah. Yeah.
Yeah. Yeah. Yeah. Yeah.
Yeah. Stuff's happening today or speculation about the future which is really what we're thinking about. Try to connect the two but we're thinking about the future. So, any of that, get in touch.
And if you found that this show or one of the other shows has gotten under your skin. Oh. That's a bad humor, okay. Yeah.
Well, until next time, keep your wafting up. Keep your wafting up. Okay. Oh.
Oh boy. It's. Oh boy.