We're Building Beings We Can't Control – And It's Happening Now

Trust Factor with Rob Greenlee focuses on all aspects to building human trust in online communications.

Well, thank you so much for being here for Trust Factor. This program is really all about the essential elements and, and the profound impacts on your happiness, security, safety, and future. And who or what do you trust as we look to our present and as you look to our future? And ahead in the show today, we're going to do a deep, deep dive on the future of trust in the world of AI, AGI, and autonomous robots., and I've got a terrific guest with me, and the goal of the show really is to help us understand the trust in our lives, and being that we're being pushed to the limit in our perceptions around the impacts of AI and the future of AGI, I've got a terrific guest with me. His name is Robert Scoble, and I'll bring him up on the stage here a second. He's probably most known for being the Scobleizer at scobalizer.com, and I'll pull up his website here in a minute. Uh, he's also a host of an online video show that's on YouTube as well as X called Unaligned. I happen to be wearing his— a shirt from his brand. So, so there's— it's kind of a cool brand as it comes to AI and robots and things like that, how, how we're all struggling to cope and and really kind of deal with this kind of feeling of being unaligned with the technology and how it's, how it's developing, you know, developing as we look to the future. So, uh, Robert is a renowned technology evangelist. He's a futurist and known for his deep insights into emerging technologies, and that includes AI, AR, VR, and cloud computing. He's, he's been around the tech landscape for a long time, is very well known in Silicon Valley and around the world for being an outspoken evangelist and really a passionate kind of aggregator of thought leadership in the technology sector and has really been known for many years for his blog. And so let's go ahead and bring Robert up to the stage, and I'll play a little bit of a a video here in a second from his Unaligned TV show, and, and so we can really see what he's up to. So let's bring him up to the stage. And Robert, thank you for joining me.

Robert, what's going on? You know, trust when it comes to technology is about to really change. I, I trust, uh, my family's life, uh, in an autonomous car next to a wall going 80 miles an hour.

I do the same thing. I just drove all the way down to Washington, DC with, um, autonomous driving, and it took me all the way down there.

Yeah, yeah. So, ah, we've been around. We're getting old together, man. I was thinking about— I was Apple's first child laborer, and Siri was launched on my show, and I've just had a crazy, uh, career of watching the bleeding edge of the tech industry. And oh yeah, we're about to change into a different kind of technology. I mean, in 5 years, there's a good chance you and I have a, a robot in our house.

I've been thinking a lot about that, actually. I, I, I think I have a robot in my house right now, and it's called a Tesla Model 3.

That's true, but it's in the garage and can't come in the kitchen and do your dishes for you.

That's— well, that's true, But I think that there is some things that the robot can do that may be a little bit more value packed.

Yeah.

Right? Right.

But are we gonna trust it, you know, with our house? And are we gonna let it, you know, hey, humanoid robot, go clean all my drawers and it finds the sex toys next to the bed. That's another level of trust, right?

Well, and also, you know—

Most people are not ready for that, you know?

Yeah, it's about privacy too. I mean—

That's what I mean. It's gonna have cameras on its eyes and it, You know, and it knows everything about the world around you, and it probably knows what kind of sex toys you have and, you know, how to buy another one on amazon.com, right? Because amazon.com sells sex toys.

So yeah, and you've been known for a long time of kind of trying to pull back the, the covers on this, right? Peel back the onion, or however you want to say it, on the— what's been termed many times by you is the freaky factor, right?

Yeah.

And that's kind of what, uh, what we're coming into is an age of lots of freaky factors.

Well, here, here's one. Last Saturday, an entrepreneur came to my house and showed me a prototype of what he was building. He's building a virtual being that uses a cell phone camera to see. So I was walking around with my camera and a virtual being was in my glasses. It could see the kinds of lights I had. In my house, right? And go, oh, nice podcasting setup. You know, I'm like, what? My— an AI know— can tell me about the, you know, teleprompter I have and the lights I have and the microphone I have, right? And, and judge me. It's like, oh, there's a lot that's about to change.

Yeah. And I do agree. Well, you know, let me step back and say a little bit that part of this is about sensors, right? Getting into our lives and being able to detect data in the real world and be able to feed that up to some sort of algorithm or bigger kind of tech stack that can analyze that and make recommendations to us, actually make decisions for us. But that's a whole element.

It knows everything in the Smithsonian. I walked around Washington, DC recently with my kids and took pictures like in the Air and Space Museum, the Holocaust Museum, and went to Thomas Jefferson's house, took pictures there, and I fed it into ChatGPT from OpenAI company, and it knew everything.

Everything. That's, that's freaky, right? Everything.

It knew everything. It's like everything I took a picture of, it could tell me what, what, where I was standing, what I was looking at, you know, all sorts of things. It's like, whoa, people. This is what I'm saying. When you get a humanoid robot, it's going to know everything in your house, and you're gonna go, how does it know every every, every brand in your house because it knows already, you know, certainly knows what Coca-Cola is, right?

And I think that's been what one of the kind of dark spots of the, of the data gathering, right, is, is kind of local neighborhoods and, you know, in your own personal residence. Um, those kind of things have been a little bit of a dark spot for AI and technology in general to actually know those things, you know. Granted, we're, we're, we're putting lots of photographs out there and lots of other types of information about ourselves. But, but, you know, I agree with you. I think that everything is going to be under surveillance or able to be quantified in some sort of a sensor. You know, even our human bodies can be detected at a much deeper level. And I think there is aspirations for getting inside the body and really understanding what's going on. And that could be good, or it could be kind of scary too.

Dag Kittlaus, uh, I don't know where my my phone went, but it's outside. Doug Kitlaus, the co-founder of Siri, one of the 3 co-founders of Siri, just started a new company called Reva Health. And he showed me what he's building. You put your finger on an iPhone camera and it tells you your blood pressure. So you're absolutely right. We're about to get a lot of new capabilities because of AI and these new large language models that are running like ChatGPT or Google Gemini or Anthropic Claude or Perplexity, these kinds of things. And they're getting vision feature, and they're getting voice features at a frightening rate, and they're getting better, right? So you start going, oh, she's— the new, uh, ChatGPT mobile app, you talk to it and it talks back in real time, right? This is not Siri anymore. This is a new, new form of AI that's doing this, and it knows everything that you aim the camera at, right? So you're at Yosemite National Park, it aim the camera, it'll tell you, oh, that's Half Dome over there, right? That's Yosemite Falls. It'll tell you that because it knows everything. It knows everything. Let's just start with that kind of understanding of where we're about to go.

Yeah. And then also layer on top of that, Robert, things that you've been talking about— augmented reality too. So how do these— out of those two things, and then we'll dive into the whole kind of, kind of robot stuff, but How is augmented reality going to factor into this?

We're about to get a pair of AI-driven glasses from a number of different vendors. I just was at Display Week a couple months ago here in Silicon Valley, and I saw the new augmented reality glasses that are coming out. They have a projector that's a quarter the size of a sugar cube at the edge of a waveguide that looks like, you know, something like this. And there's a projector shoving light into the glass, and then there's trillions of little, or billions of little mirrors that are reflecting that image into your eye. And it gives you a 720p monitor in front of you at about a 70-degree field of view, which means the glasses aren't going to be completely immersive. I mean, I, I just was at Somnium Space in Prague, and they have a VR headset that's 140-degree field of view, so twice as much field of view, and you're really immersed. Like, if you're flying a flight flight simulator in that. It's really immersive and really messes with you, with your mind, and into thinking that you're in inside a real, uh, plane flying around.

Yeah, yeah, it's, it's, it's gonna be really—

glasses are gonna be pretty fun to play with when they come out. There's several of them coming in the next 18 months. The Snap is going to announce one this fall. Mark Zuckerberg's, uh, Meta is coming out with one that supposedly they're going to show it off this fall for the first time and then have it for sale in about 18 months.

So is the form factor a little closer to what you would have with eyeglasses? Is that the model?

Yeah, lighter weight displays. Yeah, these waveguides that I'm talking— so the augmented reality world is going to split into two. One is going to be— let me get my Vision Pro. One is a heavier device like this, right? The Vision Pro, a bigger, heavier device that has 4K displays inside and has more immersion. It gives you more field of view so that you feel like you're really in the thing. The glasses I'm talking about are 720p displays and less field of view, about 70— let's call it 60 or 70 degrees field of view, which is not completely immersive, but pretty good, right? Better than the HoloLens and a lightweight pair of glasses. And that'll kick off I, you know, it's not that the mainstream is going to show up in 18 months, but you and I are going to get a pair in 18 months, and that'll kick off the mainstream over the next 3, 4 years.

Yeah, as I was kind of doing a little deeper research on this topic of where we're moving with AI and robotics and stuff, I, I just found so much information about how this is, how this could roll out and how it's going to potentially impact our world and how we think about technology and all of our work and our lives. And I mean, it's such a profound shift that that's a big reason why I wanted to do this episode was there's just so many elements to it. And I think we could probably do a 3-hour episode on it because there's just so much implication.

Yeah, every, every night or two I'm on Twitter or an X space, an audio space, talking to people from around the world about, about this new world coming at us. And yeah, we can talk for hours, many hours. Yeah.

And it probably wouldn't even cover most of the situation. Yeah.

I mean, but certainly you can cover it superficially. But if you are an engineer and building this stuff, we're not going to get very deep, you know, we're not going to tell you how to take a sensor reading off the eyes and do something with the hands, you know, that takes a little bit more, uh, talent than I have.

So I wanted to pull up your, your longtime blog just so people could, uh, see it.

And yeah, I'm wearing the Tesla shirt, that's good.

And, and this will kind of kick off our a bigger topic. And your, um, your website is here. Let me pull it up here. Um, I've got it as far as your, your address. I thought I had it here handy. Let's see here. Um, it's in here somewhere. Where did it go? All right, maybe I don't, but I thought I did. So let's see here so I can find it here. Oh yeah, here it is. So yeah, so here's your two, you know, actually the first two domain names are, are your website. So blog is the best place to go, and you've got really long-form content over there.

So, and on X, I'd write every once in a while and quirk a couple thousand words.

Yeah, so yeah, so it's awesome. And this article really goes into a lot of detail about this topic, about the Tesla Optimus and the opportunity of that and where it's going in its consumer strategy as you look out to 2028 and beyond. And I think, I don't know what you think the timeframes on these things are, but it does seem like it's coming.

Yeah, there's several humanoid robots coming next year. 1X.Tech told me they're coming next year. And the AI team told me this stuff is happening a lot faster than people know about. So, okay, yeah. Now, you know, when is it going to be that, you know, all my family and friends has a robot in their house? We're still talking probably the 2030s come, you know, for that. But me and you, I mean, I— there's a Palo Alto company down here that's making a little 4-foot humanoid kind of thing. They're 3D printing it, and it costs about $8,000 $45,000 to build one of those on a 3D printer and, and buy the electric motors and everything for it. So it shows that the costs are coming down to the consumer level. It's still a little expensive for a consumer, you know, but we can see how we're going to get there. I mean, Steve Wozniak showed me his first color printer back in 1989. In fact, it was probably the first color printer in Silicon Valley. It was a dye sublimation printer to run Photoshop 1.0, and it cost $45,000. And today my $70 printer way better, right? So it went from $45,000 to $70 in 20 years or so.

Yeah. So, uh, what do you think the, the costs on these things are going to be when they start to become available?

Right. There's so much demand for, for using it in labor, kind of, you know. My brother wants to buy 3 of them. He has a— he owns a bar, you know, he owns 2 bars in Virginia, and he's like, I'd buy, I'd buy a couple of those right away, and I'd pay $100,000 for them, right? And so, because there's so much need for, for labor like that, right? Yeah, that's true. You know, and if it gets that good where it could wash dishes or load up a beer keg or, you know, do menial things, it really can be useful in a business like that. And, and it's, you know, cheaper than buying— than hiring another employee for $20 an hour.

Right? That's certainly true. And it doesn't have to sleep. I guess it only has to have some time to charge.

Yeah, or you could get two of them and, you know, one's charging, one's working.

Yeah, and it's constantly going, right?

But now you— or maybe there's a battery pack, it'll unplug it and put it on a charger and take the fresh one off and plug it into itself and go back to work. You know, maybe it'll have a hot swap kind of something like that, right?

Yeah, that's certainly possible. I know that, yeah, Tesla's been working on this, uh, a robotic arm that can come down and plug into your Tesla, uh, when you pull in the garage or at a charging station, so you don't even have to get out of the car to plug your car in.

In a couple years, there's not going to be a human in the car, right? So it needs to have a robot like that, that'll plug you in. 'Cause someday you're gonna send your car to go get charged or it'll just automatically, oh, I'm down to 20%, I better go get charged. And yeah, how is it gonna get plugged in there? Is there gonna be a human there? I don't think so. It's gonna be a robot.

Yeah, and a lot of people are going to be able to make their Teslas available for the Robotaxi fleet. But I thought it was interesting though that the Uber CEO came out and made some, critical comments about the, the plans for Tesla on their, their robotaxi fleet, right? That would be from regular Tesla owners, right? And I thought he raised some kind of interesting questions about that. Is, are, are people generally going to want to let their cars be rode in by random strangers? I think it's a big question.

It is, but there's a lot of other things that will add value to the car. Long before you start picking up strangers. For instance, on Friday, I'm, uh, or on Saturday, I have to drop my wife off at the airport. Well, you know, she, she often takes an Uber, so why not take a Tesla? Why not take our Tesla and have it drive back to me, right? Why, why are you parking a car in a parking garage and paying $35 a day for the parking garage when it could just turn around and go park at home? Yeah, yeah, yeah.

There's certainly a lot of, a lot of scenarios that are going to come up about how to, how to use these, these robots of sorts, if they're cars or, you know, humanoid robots, that it, it's hard to imagine. And those are things that, um, that will come up as we get these type, type of, type of devices. But, you know, like you just expressed, I, I hadn't thought of that part, but I have thought about you know, having someone take me to the airport and then just ride back in the car, it's not like it's a big burden to them because they don't really have to drive. They just have to turn it on and then come back. Right. So it goes both ways. But certainly, you know, having it not need a human in the car is a whole interesting kind of deal. And then it goes back to where it's stored.

Because you could, you could send your, your car to pick up a family member, or you go over and pick up your best friend and bring your best friend over the house, right? Yeah. Um, yeah, stuff like that, right? Yeah. And certainly if you're gonna do anything like drinking or anything, you want an autonomous car because then you can, you know, not worry about causing a tragedy or getting arrested or something like that, right?

Yeah, I think they call that drunk mode, right? Is that the— is that what they're gonna—

well, you know, and we're talking a couple years, a few years, but you can already do that in San Francisco. Mexico, you could take— if you're going out to the bars, you can take a Waymo back to the hotel, right? Right, exactly. There's no human in the car, it's already here, right? So, um, it's not just Tesla doing this. There's Zoox and Waymo and Cruise from General Motors, which just made a deal with Uber just yesterday. They announced that they're gonna supply autonomous cars to the Uber network, which is real smart of General Motors to do.

Right. Yeah, yeah, exactly. Yeah. So as we move closer to the future of AI, I'm going to transition into talking a little bit more about AI and robots. But, um, you know, the, the artificial general intelligence— I, I've been hearing talk. I heard, um, I think it was Peter Thiel talk about this here recently on, on the whole Turing skills, right, as a, as a bigger factor in AI than probably AGI, right? Yeah. It feels like AGI is a little bit of a, of a too, you know, it's too ethereal of what it really means. But, but I think—

well, here's, here's a simple way to think of it. When I went to the 1X robot company, they spilled some coffee on the table on purpose to watch how the robot would clean it up. And the robot had to be told to clean it up. Hey, there's a mess here. Can you clean it up? With future AIs, it's going to do that automatically. Right. Right. Because it's going to be smarter. Right now, it can't really take its own actions. It has to wait for directions from a human being. But with the AGI world, it's going to be smart enough to take its own actions and plan out its own day and plan out its own tasks.

Right.

Yeah. And that, to me, that's, that's when AGI is really here. And, and many people in the industry say it's, you know, 2 years away, which— 18 months again. Yeah.

Right. Yeah. So this, this Turing test, I, I guess, is really more about kind of, uh, the AI having human characteristics in, in their their ability to process information and make decisions, right? And they're, they're more autonomous, I guess, and taking less direction. So we're probably going to have, what, two different types of robots? Ones that are very much, um, task-driven, or, or— well, you got your vacuuming to do, and then ones that think for themselves, right?

Well, you got your vacuuming robot, which its mission is to keep your floor clean, right? And that's all it really does. It's very specialized kind of robot. This humanoid robot can do all sorts of things from, you teaching you to dance to washing your dishes, right?

Yeah, yeah. So there's going to be probably two major types of, of these AI-driven robots, right, that have, uh, different skills or abilities. And you'll probably pay a premium for an AGI robot as opposed to a, to a skills robot that we see right now in manufacturing, and ones that are very kind of task-oriented.

So I wanted to mention also the AI is getting smarter, so you're going to even see the Even the vacuuming robots are really getting much more capable because they— first of all, they're getting camera sensors and 3D sensors on these little robots, which they didn't have before, so they can see things like wires or couches or, you know, stuff to, to go and clean and, and avoid, right? Fringes. They can understand that a lot better. Um, and the, the software, the AI itself is just getting much more capable of understanding the world it's in. And so it just does a much better job and it has new capabilities. But yeah, I was thinking back to your, your show name, Trust. You know, there's a lot that's kind of come with the robot because it— a humanoid robot is going to know everything about you.

Literally. Thus, you know, it's going to be connected to the internet. It's going to be connected to AI models, which are learning, right? So, yeah, so there it is, right?

And, you know, like, like, it'll know, you know, how to find you a specific thing in a Lowe's, right? On what aisle and what— where is it? And once we get enough data, it'll be really accurate. Right now it's sort of accurate, right? Yeah, it's—

yeah, yeah, exactly.

You know, and a humanoid robot needs to know where a screw is in a Lowe's because you're gonna send it to the Lowe's to get you a screwdriver or screw, you know. And so it better be able to find that in the Lowe's. It already does, but you know, this is the world model that's being built. And right, you know, that's a lot of trust to give a robot because it's going to know Everything. It's great. I mean, I, I'm joking around about the sex toys, but not really. I mean, you're going to tell your robot to clean all your drawers one day and it's going to find all your sex toys. Yeah, if you have any. Yeah, right. That's a, that's a lot of trust for people to give a computer, you know. Yeah.

And I still see some big, uh, holes even in the Tesla, um, Full Self-Driving. In its ability to navigate the world. Yeah, I was thinking about this yesterday when I was coming back from Washington, D.C., is that I'm in like a parking lot at like a retail store or a restaurant or something like that. And I activate the self-driving. It currently still struggles to get out of the parking lot. And it's really because I'm not sure that the maps out there like Google Maps and those kind of things are really at the detail level. Most of them are just mapping the main roads, right? So, so the Tesla vehicle doesn't really have good access. So I would think that the, the Google Maps are going to have to be more complete to the, to the full landscape and be able to index everything that's totally surface, right?

Keep in mind, I met the mapping team at Tesla one time and they said Everything you see is Google. Google makes the tiles on the display of the map, but everything underneath that is Tesla's, right? All the data is Tesla's. And so they could rip out the Google part and put in their own part and get a much more capable map for their users. And I'm expecting something like that in the next 18 months because the data Tesla has is It's incredible. A Tesla goes across the Golden Gate Bridge every 3 seconds. Nobody else does. Nobody else has an AI computer going across the Golden Gate Bridge at that kind of quantity, right? That's right. So you start thinking about, oh, wait a second, it has 9 cameras, and we have multimodal models now that can tell everything. So if it goes by a, a truck on fire on the bridge, it can tell God, you know, up in the sky, the cloud computing up in the sky for Tesla, hey, there's a truck on fire. Here's a, here's a Gaussian splat of the truck burning down, right? Yeah. And then people could be using glasses and actually walk around the truck on fire in digital, digital truck, right? You know, right. And that's a new kind of map because it needs to put that on the map and, and explain where you should, you know, where the computer should take you, right? And avoid, because there's a police action all of a sudden because there's a truck burning down, and it can tell you, oh, there's 3 lanes of traffic blocked, you know, get over to lane 1. Yeah, and you'll see all these Teslas all of a sudden just changing lanes to lane 1, and you'll be like, what the hell is going on? And then you'll— a mile later, you'll see a truck on fire, and you're like, how did they all know that there's a truck on fire in lane 3 and they all moved over lane 1? How the hell was— how the heck did that happen?

Yeah, it was because prior Teslas went through there, saw it, right?

Every 3 seconds, right?

Yeah. Yeah. So yeah, it's interesting.

It can know how long it took the fire truck to arrive to start putting out the fire. Right. So now it can feed that data back to the, to the mayor and say, hey, your firefighters are taking 15 minutes to get to the, to the fire. They're going slow. Need some motivation or a new system or a new fire, you know, they could, right, spend some money and build a new firehouse that's closer to the thing.

Yeah, so I think that the big takeaway here is, is of this conversation right now is, is that we have just seen kind of just a fraction of the amount of data gathering about our our world, right? And it's gonna be mapped and it's gonna be put into the network and these full self-driving software. And a lot of people probably don't know this, but that same software is being put into, or that same kind of self-driving neural network is gonna be put into the Optimus robot, right? Or it's already in there. I'm sure that that software is—

and your augmented reality glasses too, by the way. When you do get a robot, you're gonna probably need a pair of glasses to, to control the robot, because one, one thing about that's magical is you can embody yourself in a robot, right? If you have a humanoid robot and you have, you know, a Vision Pro or something like that, you can actually put yourself into the robot and walk around as a robot. It's crazy to think about that, but I already did it at the, at the 1X Tech, uh, robot company, and it's really crazy.

So I hadn't thought about that before, but I— you're exactly right. And that kind of changes the dynamic, doesn't it, about human robot control? Yeah.

And also, if you're— let's say you're trying to learn kickboxing from your robot, you're fighting your robot, which probably is going to be a few years because You don't want to be hit by these things. They're going to hurt.

Steel hands are not going to be a good thing, right?

No, the 3D-printed carbon fiber that they're doing at K-Scale, which is the robot company in Palo Alto. Yeah. I banged on it with a hammer and I didn't even leave a mark. I not just didn't break it, didn't even leave a mark with a hammer., right? So imagine that thing hitting you. That's going to hurt. And you're not going to hurt it, right?

It doesn't have any— well, it doesn't yet have feelings, but it certainly doesn't have a real body per se, right?

It knows about like pain, but it just, you know, it doesn't feel pain the same way we do, right? We're biological beings. That's a digital being, but it can know that it did something wrong, like put its hand on a hot stove.

Yeah. Yeah, that's interesting to think about that. Now, one thing that everybody's worried about right now is job losses as a result of robots. And I just wanted to throw some numbers out that I found is that there are people out there and tell me, Robert, if these are wrong or from what you've heard, but they're forecasting job losses about 20 million manufacturing jobs. Yeah. By 2030, and about 30% of all jobs will be automated by 2030.

So it isn't like it's happening like next year or something like that, but it's already starting to happen. I know of call centers that have gone completely automated, laid off a few hundred people. I met a guy at a game developer conference who said they just turned on their first automated cold storage warehouse. So there's only robots inside the warehouse moving salad around and stuff like that. And I keep hearing these kinds of stories. So I know it's starting to have an effect. And, you know, I've been talking to the auto industry because I'm rewriting my spatial computing book with Irina Kronin. And in China, they have completely automated car factories, no humans, supposedly. You know, I'm sure there's still a few humans walking around here and there, but the factory itself is all automated. And you talk to the other, You know, you, you go to Detroit and you say, hey, when are you going to turn on your first fully automated factory? Well, they can't do that because of the labor unions. Oh yeah, right. I imagine going to the labor union boss going, hey, we're, uh, we gotta, we gotta open up a multi-billion dollar completely automated factory. We're not going to hire you guys anymore. Sorry.

Well, that's, that's a tough one, but in China they've already done That day's coming, isn't it? Yeah, yeah, there's no question. Yeah, and I also wanted to talk a little bit about, um, all the different types of robots. I think this isn't a, um, a topic that I hear talked about a lot, um, but, but, you know, kind of the— yeah, the type of robots and kind of their, their bigger use case. So like, uh, robots for healthcare, um, and, you know, or washing windows.

The Chinese have a little robot that rolls around the piece of glass and cleans it, right? But But the problem is with that little specialized robot, I have like 20 pieces of glass in my house. So I have to move the robot 20 times to clean all the glass in my house, right? Probably twice. So 40 times inside too, right? And it's a pain in the butt. And sometimes you can't reach the thing or whatever. Well, the humanoid robot can do that drungy work of moving the specialized robot around the glass and clean all your glass. And it'll charge you $40 to clean all the glass in your house, right?

Yeah, I think we've been living with robots for many, many years, and I'm not sure that people really think about these robots, you know, like the Roomba that we mentioned earlier. And I think that there's been robots in healthcare for many years. You know, there's these kind of armed-based kind of surgical robots. Yeah, Yeah, surgical procedures where a doctor—

it's really cool using the, uh, surgical robot. The haptics on that thing are just sick, you know, because you— if you put a needle through a fake piece of skin, you— it actually— you feel it going through the skin. There's no feeling, it's all digital. The robot's across the room, there's no hookup, physical hookup to this, to the surgery tools, right? But it's all, uh, using haptics.

It's really amazing technology. Yeah, so, so I think robots have been a part of our lives for many, many years now. It's just that maybe we haven't thought of them in that context. They, you know, the Roomba has been more identified as a, as a robot, but, but I'm not sure that these other, you know—

well, we don't get to see surgical robots very often, right? No, we don't, right? Unless you're, unless you're a surgeon, you see them all the, all the time. But as a normal everyday person, you you probably haven't had your hand on a surgical robot, right?

Right, right. But, you know, these humanized robots could, could replace nurses. They could, could replace surgeons at some point. They could replace doctors entirely. Yeah, someday, you know, someday in the future.

There's a lot more empathy upgrades before it could do that, right?

Right. But I think I think, you know, as we talked about earlier, it certainly helped out, you know, right?

There's a lot of things to do in a hospital that don't require a human touch, right?

Right.

Yeah, like physical therapy, I think delivering supplies to patients, and even bringing food to people, you know, that are in a hospital bed, or moving them a little bit so they don't get a bed sore, right? A couple robots will come in and, hey, it's time to move you around a little bit.

Right? Yep, yep. And that will probably save the hospital, you know, money.

I mean, and it'll be better for the patient because, you know, who wants a human being to come in your room and see you naked and move you around?

Yeah, that's right.

That's true. I'd rather have a robot do that kind of work, you know.

Yeah, exactly. And then, then I think the home life side. So we had talked about earlier about dishes and cleaning the house and mowing the lawn and those kind of activities.

The lawn is probably going to be a specialized robot, right?

If you really— well, we have those right now, but they look like they're, they're like four-wheel kind of rolling machines, right? They don't really have a human form factor to them.

The, the current ones, uh, you have to put a metal wire around your lawn so it knows the edge of the lawn. The new ones, like Empire Sheep Company, I went and visited them. They're making robots to do landscaping, and their robot can get on a piece of lawn without the wire, right? It knows, it figures out where the lawn is just from the cameras.

Yeah, yeah, yeah. I think that that's where it's going, right? Just like the Teslas trying to look at the world like humans do and navigate it and make decisions and things like that. That's, that's where we're going. So these, these limit wires that have been around, or the dog restriction area kind of devices are going to be replaced by vision.

Yeah.

So it's going to be interesting to see how that plays out over time. And then we talked about the manufacturing robots and, and I think that's pretty obvious that that's going to be, you know, I mean, you just look at an Amazon warehouse and they probably have, you know, millions of robots already. So, you know, this isn't a new thing, but, but if I was a warehouse worker in an Amazon warehouse, I would be a little concerned about my future job.

So unless you fix robots, and then you probably have a job for a while until they start fixing themselves, you know. Robots, man, somebody has to fix those things. Every, every once in a while they break, you know.

Right. Yeah, I was hoping to talk about that, is that, you know, what's the changing landscape for, for work and, and jobs in this environment, right? I do think, you know, just like what you said, managing the robots or—

We do, right. We do need a new policy and we do need to have, you know, a debate, discussion about, about the policy. Is it, let's say we take a truck driver off of, off of driving a truck because that'll happen in the next 15 years, right? What do we do with that truck driver? They were making $80 grand, $100, $120, $80,000 to $120,000 a year. And all of a sudden they're not able to feed their family. That causes not good things in our society, right? So we need to figure out how do we— do we have a guaranteed minimum income for that person that they just sit at home and collect a paycheck? So, you know, or do we, uh, give them the wealth to go back to school and spend 4 years and get, get a degree so they can repair robots or something? There's nothing new. Right, right, right. Yeah. And, or even, you know, teach them how to do plumbing or put glass in. My brother runs a glass shop, right? My brother-in-law runs a glass shop. And so he used to be a tech executive, and he learned over 4 years how to do the glass and buildings and stuff like that. It took, you know, a few years, he told me, to pass all the licensing and learn all, all the things that it takes to learn to run a glass shop, right? And so you need to take a truck driver and, and give him the wealth to spend 4 years learning a new thing, right? To have a shot at another American dream. And so we need to have a policy. My dad escaped poverty in New York by going to the Army and then getting the wealth to go to school and getting a PhD in electrical and material science, right? And that pulled us all out of poverty. We— before that, our family was living in the projects in Brooklyn, right? So, you know, same thing is needed for everybody who gets laid off. They need a new American dream. They need a new shot. And so we need a new policy. But how do you sell that to Americans? Because we're so divided politically. That we can't even talk to each other.

Yeah, well, and there's two, two completely different visions. One is kind of looking at how, how the world has been, uh, as far as our economic structure, work culture, what our, what our beliefs are, and then thinking about this new technology-based kind of change that's coming and and how that's going to impact people's lives. And it's a, it's, it's a huge transition. And I am a little bit worried that this, this transition to this new kind of world that we're coming into of, of, you know, I hate to say, but surveillance and data gathering and aggregation of, of, of data to kind of fuel the data that's needed for these robots and the technological innovations that we're right on the cusp of, that that's going to cause a lot of people to have a lot of change anxiety. And I think we're starting to see the beginning of it. I'm just not sure how long this transition is going to take and what the costs are along that transition.

A while. I mean, to take a human— I know humans who have gone from cooking in a kitchen to running a computer, uh, company, but it took a very special school to do that and took a few years. Right? Because you, you can't just learn how to program a computer if you're a truck driver. Learning how to program a computer is hard for that person, right? And so you got to go to school and, and have the time and not have to work 3, 3 jobs just to put the money on your table to feed your family, right? You can't be working an Uber job while going to school. You're not going to have time to learn. Right. Yeah, yeah, it's true. I know people do it and, and it can be done, but it's far harder to learn, to really learn something new if you're working at, you know, two jobs during the day, and then you get home and you're tired and you're not doing your homework properly.

Yeah, I mean, I can see a day when, when one person, one human owns, let's say, 1,000 robots and they, they can build a company, that, that offers services, whether it be plumbing or, you know, lawn mowing or, you know, window installation with their 1,000 robots, and they've built maybe a, you know, a $10 or $12 or $15 billion or million dollar company, uh, just based on having all their workforce be robots, right? Yeah. Um, is that kind of what the future of work looks like for us is is who owns these, right? And is there going to be robots that nobody owns?

Ooh, like, like a government owns, or that are completely ownerless? That's an interesting concept to think about. Like, who would make it? Why, why would it be made? Why would it be put in service, you know?

And, and what if, what if robots made robots?

Yeah, but then the robot owns it, right? And who owns the robot? Elon Musk. You know, Elon's going to claim ownership. If an Optimus makes another Optimus, who's going to own the new Optimus? You know, it's Elon.

You know, Elon's getting high on that thing. Yeah, I thought that the whole mantra as we look into the future was, is that you will own nothing and be happy. So, so maybe there's, there's a scenario that, you know, maybe these things become sentient autonomous beings at some point. I mean, this is really far off in the future, right?

Yeah. Give us 20 to 50 years.

Weird shit like that happens. That's really, really going into the freaky zone to start thinking about that, right?

Are you— here's why. I know you're going to bring your robot on vacation with you. Let's say you go to Europe for a month on vacation. You got to bring your robot with you. I'm going to bring mine because it's very useful.

Right? Yeah, but why would you bring it is the question, right?

I might rent another one over in Europe and call that mine, right? And for the month. And but I'm going to have a humanoid robot with me on vacation because if I want— let's, let's say I'm in New York and I want to go to a pizza place. Well, you got to stand in line for an hour sometimes at some of those places. So I'm going to have the robot stand in line for me and I'm going to go and, you know, sit at the coffee shop across the street until the robot gets near the front of the line. Then I'll pop in with the robot.

So I do know that there's a lot of—

Or even go further, just have the robot get what you need, you know, the pizza and bring it over to you.

That's a true vacation. You can just kick back and the robot can do all of the grunt work. While you're on it, right? Right. Yeah. So I also know in Japan that there's a lot of luggage, you know, a lot of, um, robot research being done around, um, robots that could be like companions or, you know, relationship robots, those kind of things that, that maybe are female, uh, males, you know, going back and forth, and humans get into the bedroom.

Off the table. All right, robot doesn't go in the bedroom. Room. At least not for that purpose. Robot will go in and clean your bedroom, organize your bedroom.

But yeah, I'm not insinuating anything. I know Elon has joked about this, right?

No, but you're gonna fall in love with the robot in the other places because it's going to make your grandma's favorite recipe for you whenever you want, or clean your toilets, or, or play chess with you, right? You're gonna be like, yeah, fuck, this is really cool.

Right. So there's probably a future where humans get into relationships with these robots, especially as they become—

Already happened. CNN had a show 8 years ago about people who fell in love with robots 8 years ago.

And they were pretty, they're pretty ugly and pretty stupid robots compared to what's coming.

So jerky and right.

No.

Yeah. They don't look like a human. They don't talk to you. Right. The AI just wasn't ready until just right now, right? So to do really amazing robots, so I know people are going to fall in love with the robot, even if they leave sex off the table, right? Yeah.

Well, because it's going to, it's going to be pretty fun to have a robot. Yeah. But you have to admit that, you know, with some of them, that's going to come as a, as part of the package, right?

So. You know, extra fee. You know, there's, there'll be some robots that'll be like, no problem, let's go.

Because you and I both know that there's going to be big money in that, right?

So it's a weird future, man.

Yeah, yeah, it certainly is. And then, you know, one, a couple areas we haven't really talked about yet too is kind of like space exploration. And the, the biggest fear is military applications. Yeah. For these robots, right?

Yeah.

You know, I, I don't know about you, but I've had this thought that likely if Elon Musk is going to send the Starship to Mars at some point, I think he stated that he's shooting for like, what, 2026 or 2028 or something like that, that he's shooting for.

You see the engine designs and it's like, okay, it's coming.

Yeah. Yeah, I mean, why wouldn't he send his first mission of autonomous robots to Mars, not humans?

I would recommend that because there's a lot of radiation pelting the surface of Mars. And unless you have— my dad is a radiation expert. He said you need 6 feet of water between you and the sky or a good chunk of lead. And both of those are really hard to do in space because it costs a lot of money to get that much water up there and/or dig it out of the ground and process it and get that much water. And it's hard to get that weight of lead up there. So you're going to live underground if we get there. But you're right. Why not send the robots to set up the tunnels and build some homes and build some sort of thing thing that would support human life before we get up there.

Yeah, yeah, I can see that. I mean, it just makes sense. I mean, he's starting to— I'm talking about Elon Musk— he's starting to roll these robots out first in the Tesla manufacturing factory, which is kind of like an obvious move, right? Because he thinks that that's the first step, is the manufacturing side.

It's who, it's who has the most demand, therefore who can pay the most for the first robots, right? Yeah, he's a profit, he's a capitalist, he wants to make money, so I'm sure he wants to take the low-hanging fruit first, which is, you know, a bunch of workshops, right?

Yeah, I would think that he would want to, to kind of cut down on his labor costs in his manufacturing facility because he's under a lot of pressure to get these Tesla prices down.

His Chinese competitor has a factory that has no fucking humans in it. Totally autonomous, right? So that's the pressure. If you don't have one of those, you're not going to be as profitable as the Chinese, and you're not going to be able to compete with the Chinese on quality score. Yeah, yeah. Well, so you want to, you want to have access to the Chinese market, you better have an autonomous factory soon.

Yeah, or something really close to that. Yeah, on the space exploration side, we've, we've had lots of robots that have been doing work on Mars and things like that.

I mean, I know that they get robots, rovers.

Yeah, right. They take a certain amount of direction from humans on Earth, but they have to operate kind of autonomously because the transmission time between Earth and Mars is too long for it to be real time.

It's not like they're on a couple of minutes, right? And so you tell it to turn, you tell it to stop because it's heading toward the cliff. Well, by the time it gets the stop command, it's already over the cliff. Cliff, right.

So it's got sensors in it that detect real-world kind of scenarios. So I think we've been playing around with robots in space for many years, you know.

And by the way, JPL invented SLAM map. So the technique of doing that SLAM map, well, now your robot has a SLAM map, right? It works the same way that the robot on Mars used to work., right? Yeah, yeah, same kind of technology.

So yeah, it's a, it's a new world. Yeah, and I think with the, the whole full self-driving kind of stack that Elon has, I mean, I think robots will play an increasing role in space exploration even beyond just, you know, trying to colonize Mars. It's probably going to be a lot of aspects of our space exploration is going to be driven by by autonomous robots, increasing.

A lot of the humans are going to be down on Earth building things in simulators and testing them out long before you build them in physical. Even if you send a whole bunch of robots, you're going to do that in a simulator first to train all the robots what, how to build a tunnel or how to, you know, unload a, you know, a lander, you know, and build things out of it. Right? So yeah, it's a, it's a crazy world coming at us. Yeah, trust. A lot of trust. You gotta trust. Are you gonna trust your family's life to, uh, melted sand? I don't know, right? Yes, you, you have a Tesla. You—

I trust it all the time next to the wall going 85 miles an hour, right? Oh, I did. I mean, I mean, you have to in order to use it, right? You have to let it, um, go right up to the point of failure, right? I mean, I do it all the time, right? I'll let it, um, I'll, I'll take it as far as I'm looking at.

I'll be honest with you, I'm looking at the phone on the freeway. I don't know what you're doing, but I'm letting the car drive, and it's driving real good, and it's driving right next to a wall at 85 miles an hour.

It could kill me at any point, right? Which is a scary thought to think about too, right?

It is, because you're right next to the wall, a foot and a half from the wall, going 85 miles an hour. Like, it could put me in the wall.

It wouldn't take much. Yeah, yeah. I mean, it raises the specter of, you know, if somebody wanted to take you out, it would be really easy to do that, wouldn't it?

Well, it's not really that easy because You would have to fuck with the— you would have to hack the Tesla in a way that, for instance, the Tesla, if it gets a direction to make a left turn, the sensors are then watching to make sure it makes a left turn. And if it makes a right turn, it knows something hacked it, right?

Because something perverted its system, right? So you saw that, that movie on Netflix that that showed all the Teslas without cars driving in, you know, crashing into each other. Yeah. Um, and, and it, it definitely got people thinking about that, about, you know, being able to capture control of all the Teslas at some point, because they, they are mostly all of them connected to the internet.

So yeah, loosely, you know, that's what people don't realize either.

The car is very decentralized. It, it self-drives.

That's good to know, right? Yeah, it works in Yellowstone National Park driving you around, right? And there's no connectivity unless you bring your own Starlink along, right?

Yeah. So it's, it's stored the, the map, right? So, and it's got vision, so it can, it can mostly, mostly navigate that.

Yeah, that works great. The map is a little wonky because it doesn't have access to the map data, but it drives just fine. It knows, you know, how to drive locally without talking to the cloud computing. It can drive without being connected, right? So yeah, that's true. It's a, it's a fairly well-thought-out system, and that's why it hasn't been hacked at any scale, right?

It's really hard to hack this system.

That's good to know.

I, I have a security—

definitely builds trust. Well, right, and that you'll lose a lot of trust if it does get hacked. You see somebody driven into the wall or 100 people lose their life in a or something like that, run through a red light because, uh, something perverted the system, right? You won't trust that anymore, and you'll be like, oh shit, I'm not gonna do that anymore.

I don't know that you'd be able to tell, um, if it was hacked or not, because I mean, I, I see—

yeah, you would, because, because human beings are patterns and machines are patterns, and if the pattern breaks, if all of a sudden 100 people get killed You know, you and I both know that we would be like doing a lot of research real quick, trying to figure out why did 100 just kill humans all at the same time? That's not, that's not a pattern that exists right now. Right now it doesn't put me in the wall, right?

Right.

That's true. The other thing that the tech industry knows, and for your show, is they know when people get get like 3 days worth of rides in an autonomous car, they change as consumers. They change their trust of computers because now you're using a computer to drive your family next to a wall.

That changes what you trust. Yeah, because I think for, for the longest time, Robert, I mean, both of us used to work at Microsoft, right? Yeah. And I think that this image of of trusting computers to be reliable in kind of life and death type situations has always been a little sketchy because there's been all these instances of blue screen of death, right? Yeah, yeah.

Right, that's been out there for a while. I'm still watching it close.

I don't know about you.

Yeah, yeah.

Is this thing gonna put me in the wall or not? So it did take a lot of trust Yeah. To actually get over that concern that, well, this is a computer and it's, it's really handling one of the most important tasks that I can imagine. Right. Which is managing my car. And I just have let it do a lot of things.

Right. So once somebody lets their Tesla drive them around town and go through stoplights and stuff like that without, you know, touching the car, don't you become much more accepting of like automatic shopping in your kitchen? Yeah, I think that I— I—

it's what the robot really is going to bring, right? I tend to think that maybe we've reached a point in technological kind of, kind of development that it is becoming more reliable and more trustworthy on, on the scale of safety. I don't know that I have the same level of trust when it comes to information and privacy.

I think that that is—

I think that's about to radically change. Oh, you think it's going to go the other way? Yeah, no, Apple's been talking about, you know, blocking access to YouTube.

No, no, no. I think humans are going to figure out they get a much better thing out of their robot if they give it more data. You're going to teach your robot everything about you. You're going to have a conversation one night with your robot and go, hey, I like pepperoni pizza, right? Because you're ordering pizza. And now from then on, it knows you like pepperoni pizza. So the next time you order pizza, it'll go, hey, do you want pepperoni pizza like you usually have, right?

Yeah, I already see that behavior in my Tesla. It, it tends to know what I'm— or project what I might be doing based on my history, right? Yeah. So if I usually go to, let's say, Costco on Saturday and I get in my car Saturday afternoon, it's— that's at the top of the list of destinations that I can select, right? So how did I know that? It learned that because of my past interactions with it.

Bingo. Yeah. So we're already there. What happens if you go to strip club every Friday?

Oh, wait a second. It's time to go to the strip club.

It'll, it'll, you know, it'll know where you're going. Exactly. So anyway, we're heading into a new world where we're going to give it a lot of data. There's going to be people who are like, give it all the data. Like I am, I'm giving it all the data. I'm recording every, like this call is being recorded, right? And there's AIs that'll watch this on YouTube and ingest it in a minute or two and write you a transcript. And then write you a summary and then write you notes and then, you know, tell you what's important. Yeah.

No, I, I, I, I couldn't agree more. It's crazy. So the next topic I want to talk about is mental health.

I think that'll happen. And last time we talked, there wasn't AI that ingested YouTube. Now there is. It's like, and it, the Lex Fridman podcast, it was like 8 hours long. Yeah. It took 2 minutes to ingest that. Wow. So it could watch the whole thing in 2 minutes, and then it has a transcript, then it has a summary, then it has notes, and then you can talk to it. It's like, oh, what else do you want to talk to, to this stupid video? You're gonna talk to the video?

Like, what else you got? It's actually— I mean, I would say that's probably a huge, um, kind of piece of content that would be good for training data, uh, for the AI, Right.

So I think it's going to change our jobs too. Podcasting, you know, now you have to think about it. You're, you're going to have an AI as an audience, right? Because many people are not going to watch podcasts anymore. They're just going to give, uh, you know, they're going to send this show to their AI and say, hey, summarize this for me and tell me what's important in it. And oh, okay. There's a really important thing that we would talk about for a minute. Take us right to that spot and let us listen to that or see it, right?

If I showed you something. Yeah, yeah, right. Oh yeah, totally. I mean, and I think we are seeing it in podcasting increasingly that all this data, and I think that there's a little bit of risk, um, those that are heavy podcast or content creators like myself about, you know, what the future looks like for the, the ownership that we have of our likeness and our our voices, right?

Yeah.

And how those could be used in the future. I know that YouTube is planning on translating all of the content on their platform into all the languages of the world and syncing that audio change to our lips and our videos, right, to make them align with each other. And so we're seeing, you know, this technology being used in a lot of different ways. But I wanted to talk a little about military applications. Because I think this is a huge area too. Uh, robots are obviously starting to be developed and, and deployed, um, in increasing ways, uh, to handle dangerous tasks in dangerous situations. And, and I don't know what you know about all this stuff. I mean, a lot of this stuff, a little bit top secret to some degree.

Um, not hugely top secret because they, you know, the bidding process for the military Uh, they, let's say, uh, they need a bunch of robots to do something on an aircraft carrier. They'll have an RFQ and somebody will bid on that. It's not hugely secret. If they have some weird-ass sensor that nobody knows about, okay, that might be secret, right? There's secrecy, but, um, pretty much you can figure out what the military's wanting just by, just by knowing the military, you know. For instance, an aircraft carrier has 4,000 people on it. It's a supply chain. When I was on an aircraft carrier, there was two other ships hooked up to our ship, uh, passing, you know, huge, uh, things full of food and stuff like that over to the— over to the aircraft carrier. And then most of the 4,000 people never get to fight, you know, never get to do the, the shit, you know, that, you know, go and blow things up kill things. Now most of them are fixing engines or cooking or, you know, doing various maintenance things, right? And all of those need to be trained. So one thing you learn if you get— go on an aircraft carrier is the whole thing is run by 18-year-old kids. The whole ship, that they drive the ship, 18-year-old kids, right? It's billions of dollars worth of taxpayer money. And a kid is driving it. Well, the kid had to learn to drive it, right? The 18-year-old had to learn, had to go to class and understand how to do it, right? Well, now with glasses on in a simulator, I can teach that kid how to do it much faster than with a book or even with a YouTube video.

So over time, we could see these ships being, uh, you know, really controlled and operated by AI or, or, or robotic processes, right?

I mean, a lot of these things, even if a human is doing the work because a human is still better at a certain job, they're going to be wearing augmented reality glasses and it's going to show them how to fix that thing and watch that they fixed it properly so that, you know, if that thing crashes in the future, there'll be proof that the guy didn't fix it right. Yeah, you know, and it'll keep him from making the mistake in the first place so the crash won't happen.

Yeah, I'm curious if, if everybody's fear is going to be realized where there's kind of soldier robots, um, coming. Yeah, I think I saw a news story somewhere that was talking about that, um, China was creating soldier robots.

You don't want these, uh, these little 4-foot things. Yeah, 3D printed. You can't destroy them with a hammer. You can't cut them. You can't get close enough to cut it with a chainsaw or something, or a, or a disc saw, right? It's not going to let you get close enough. It's going to whack you, right? And so it doesn't even need a gun. It's going to keep cutting. And if there's 1,000 of them marching down your street, you're not going to put up much resistance to them because you're not going to get very far with these things.

Right. And there's also the military robot dogs that are out there too.

Yeah, right. Enough said. I mean, you know, the robot military force is coming and it brings a lot of new ethical problems because if that robot's allowed to kill people on the street, you know, because that's what military robots are going to do. Who's in charge? Who's it? Who's authority? Who, who gets called into the president's office for making a mistake, right?

Who's accountable, right?

Especially if they're, they're autonomous, right? That's what I'm saying. If they're making decisions on their own, that's really a new, a new problem. And, you know, who's, who's going to break the rules, right? Because we could all get together and say, oh, we have new war rules, we're not going to do that in our war It only takes one country to break the rules and then everybody has to do it, right?

I do wonder though, if all of the countries kind of create these battle robots of sorts and that ultimately what we're gonna see is wars fought between robots versus wars fought between humans and robots. Oh, we'll do that for fun, right?

Football.

It could bring in an era where there's this constant war because there's no humans involved.

Um, yeah, that's sort of dystopian.

It is.

That's a little dystopian, right? Yeah. Iran-Iraq War went on and killed many, many of their kids, right? And they didn't get it. Neither side got anything out of that. Just killed a lot of their own kids. Yeah, you know, and you gotta dedicate a lot of your GDP to having a robot war.

That's sort of stupid, right? Yeah, but I'm sure it will play into the, the funding of the military-industrial complex though.

Uh, they're getting enough funding anyway. The real thing is we never, we never, uh, fight the next war. We're really good about fighting the last war, right? When Timothy McVeigh blew up the Federal Building, we were good about figuring out how to fix that, how to keep that from happening, right? And we built satellites to watch farms, to watch the nitrate being put on the farm, right? We changed all sorts of relationships with the chemistry industry. We track everything, right? And so now you don't have too many Timothy McVeighs blowing up federal buildings. But we're not planning for the next war, which is these things are going to be cheap. And a country like China or even America can make many robots. And, you know, if you see a couple thousand robots coming down your street, you're not going to give up. What are you going to do, shoot it with your shotgun?

You're not going to stop two of them, right? Well, I think we've seen war already happen, uh, with drones.

Uh, yeah, it's being used in Ukraine, right? And, and very cheaply.

Yeah, it's like much cheaper than the Lockheed Martin drone, you know, $200 drones, right?

They have a bomb attached to them. We didn't plan on that as a war. Oh, you mean there's going to be an adversary that's going to have a lot of little cheap-ass drones and they're going to blow shit up?

Oh my God, how do we stop that? Yeah, I've seen the actual, um,, you know, videos of that where the drone will actually chase the soldier and, and, and kind of crash into them and blow up, right? Yeah. And that's how they kill soldiers with the drones.

Yeah. Yeah. So it's craziness, right? When I, when I first, I, I was in Israel at a tech camp and I saw a general in the Israeli army teach me about drones, the old, the first drones. Right? They had a big-ass drone up in the sky that, you know, could target stuff down on the ground and spend a lot of money to blow up some guy. Now it's done with a $200 drone.

It's like it changes how you think about war if you're a war planner, right? Yeah. And, you know, and what other weapons of war do we not know about yet that they're working on? Could change the whole thing. I, I keep hearing us, um, concerns about nuclear war, um, and I do kind of wonder about that. I think it would be suicidal if anyone pulled the trigger on that, right?

So I think it's kind of— no, because how do you, how do you de-escalate from, you know, somebody— let's say Russia nukes something in Europe, how do we—

we're not going to stop.

No, nobody's going to stop, and we'll all be dead in, in like a half an Well, it might, it might take a couple days to ratchet through everything, but yeah, you know, it's not good. We don't want that kind of world for our kids or for ourselves, right?

We don't want to do that. So yeah, it's almost like, uh, you know, if somebody starts something like that, it's almost like they're, they're committing suicide, really, in some ways. Is anybody really going to pull the trigger on that going forward?

Uh, you know, hopefully nobody feels, uh, this is the problem with escalation, right? And how do you not escalate? Right now we're learning about that with Russia. You know, we'd like to go kick Russia's ass, but we can't. They got nuclear. You gotta de-escalate. You gotta take whatever they're doing and let them do it because they're nuclear power, you know. And they got to do the same thing with the US. So if we want to attack something that they like, you know, They gotta let us do it.

Yeah, it's fascinating that we've mostly maintained kind of more traditional warfare, right, with soldiers and airplanes and tanks and that kind of stuff is what— I mean, our technology is so far beyond that, it's just, it's a little—

Even with planes though, I met the F-35 fighter jet pilots down in Las Vegas, right? And one of them was like, oh, I'm going to beat any F-18 or F-16, which is what we sell to the rest of the world. And I'm like, oh, how do you know that? You know, maybe they're better talented at more talented than you are. And he's like, no, that's not how this works. Because we're flying at night and I can see you. You can't see me. I'm stealth. You're not. Start with that. Right. And then he said, I can stop in midair and you can't stop. And he said, soon, We're going to replace the pilot because the pilot, you know, if you're flying like a curve like this, you can only do 8 G's. Otherwise your head explodes. Right. Well, if a piece of melted sand is flying your plane, they don't have that problem. Right. Right. So you can do some really crazy dogfight move that a human can't do.

And you're gonna win. Yeah, so humans are generally probably not going to be casualties much as you look at the future of warfare, right? Unless they're like thermonuclear, or if there's humans on a, like, a battleship or on a, on a control craft, right?

Maybe. But maybe you have a drone army in your, in your neighborhood that it's killing people, right? I, you That's, that's the problem is what is the next war is really hard to think through because number one, we can't really talk about it on public airwaves. You know, let's plan out some terrorist attacks and see what, you know, how those would be defended.

Oh, that's not probably not a smart idea to talk about in a public space, right?

No, no, that's exactly— certain people come and visit you, hey, I give talks to those people, they're nice people, they're patriotic, but they'd come come and visit you, you know, hey, we saw you on YouTube talking about blowing up America. That's not good.

We don't like that kind of thing here. So, so it's— as we look to the future, I know that there's some predictions out there about when robots are going to become, you know, shift into a humanoid-like experience, because most of the robots Okay, it's coming up next year is when we'll see actual purchasable humanoid robots.

That's what I hear. Yeah. Okay. Now, is it good? Yeah. You know, you and I both worked at Microsoft. Was version 1 ever good? No, no, no.

It's always version 3 that's good, right?

Yeah, exactly. That's true. We used to joke about it, right? I— wait till version 3, that's, that's going to be the one you want.

Version 1 has a lot of bugs, you know. Yeah, yeah. So here's where we are right now with, um, with the Tesla Optimus robot. And so I have this up on the screen now, um, for those that are listening to this. Yeah, we're, we're, um You know, we're kind of at the stage in the demos of mostly like in manufacturing type of environments. But as you can see on the screen, the Optimus robot— and Robert, as far as you're concerned, why are we talking most about the Optimus robot? Is it just because of the FSD software ability?

Because they have the most data of the real world. They have cars driving across cross the Golden Gate Bridge every 3 seconds. Nobody else does, right? Start with that. They have more data than anybody else. They have better infrastructure for building AI technology. They have a better simulator than anybody else. They have more data. And they start with more data. More data wins, right? As long as everything else is equal. Um, and so they, they have that, and then they have the ability to manufacture things at scale, right? And you, you're like, oh, who has the— who has the ability to make a million robots?

Well, Tesla certainly does, right?

Yeah, yeah.

So do you think— certainly do, right? Do you think that Tesla is going to be the biggest robot manufacturer in the world, or is there others?

Got a rival? There's going to be rivals. Um, you know, uh, Brett Adkins' company's a rival. They got $700 million. Certainly in China You're seeing Unitree and others who are putting together pretty credible efforts. They just have the ability to build things for scale for cheaper than we could build them, right? So eventually that wins wars, right? If we really do go to war, I know how that war is going to work out. China is going to win because they're going to— they have the ability to make things faster than we can and cheaper. And that eventually wins wars. Right. Yeah. Yeah. It eventually wins Burning Man. Right. Next week, everybody's at Burning Man. You know, Steve Jurvetson, who funded Tesla and SpaceX, he has in his camp all the drone people who are doing cool drone shows. So they're having drone wars already at Burning Man for fun purposes, not for military purposes, or just to do a cool thing in the sky. Right. And there's teams competing to do cool things in the sky. Right, with swarms of drones. Yeah, sort of the same idea as a military thing, right? Yeah, yeah, it's just not cool thing in sky, it's kill you thing in sky, right? Yeah, it may be a cool thing too because the military has dual purpose. One, they, they do want the ability to kill you, but then if you have a disaster, they roll up the aircraft carrier and feed your people for a while, right? And do surgeries and stuff because your hospital was destroyed or whatever, right? So that's a humanitarian effort. The military serves humanity— humanitarian— because it wins friends, and right, wins friends around the world. And we'd rather have friends than have enemies because enemies cause wars, and we don't want wars. Right? Even if you're in the war business, you don't really want a war. I met a guy who builds, uh, nuclear bombs at Lawrence Livermore Labs, and I said, what's your job like? And he's like, I build something I hope never gets used, and I have to prove to the president that it works. Yeah, in a simulator, right? They, they're not allowed to test their product anymore. Yeah, so they have to simulate it in a simulator to make sure it will actually work if it ever is needed to work. But yeah, it's a, it's a different world, the military, right?

Because of that, right? Yeah. So let's, let's move on and talk a little about in the last, you know, 10 minutes or so of the show, um, how, how we might see these robots in kind of more of a service mode and how they, they may be monetized and what those, what those kind of processes might be, whether it's a— they're going to operate on a subscription model. I mean, if you think about your iPhone today, yeah, you know, we buy apps, right? And, or, or we acquire apps. Could be interpreted in my mind as kind of like maybe a, you know, like a skills type platform. It has certain abilities, right, in the right? Do you think that's going to be the model that is brought to, let's say, Optimus, is that you'll be able to subscribe to certain skills like the ability to mow the lawn or the ability to do dishes or things like that?

Yeah, like household chores cost $50 a month, right? Yeah. And the first one, I bet if I saw a robot on my street, the first one is going to be like like in a delivery van or an Amazon truck or something like that and replacing that, the figuring out how to get a product from my curb to my front door, right? A robot could do that. And that's part of the delivery service fee, right? We're already paying it, right? We pay a human to deliver our Amazon packages to our front door and ring the Ring doorbell, which is also owned by Amazon. So in the future, you're going to have a robot that does that, and that's going to— the fee of delivering the package is going to pay for the robot, right? So that's one thing. But if you want to bring it inside and have it set your table, maybe it'll say, hey, I, I can set your table the first time for free, but, you know, you— if you want me to come in and set the table every time, you know, starting next time, you got to pay, you know, something, $10 a month or something, right? For household food chores. Food, right? The food chore package, right? Oh, you want the food chore package? It's something like that, right? You know, so you pay your $20 a month and it'll let you set up your table and wash your dishes. It'll maybe help you cook if you, if you're doing cooking, right? Food chores. Oh, you want to clean the bathroom? That's a bathroom chore.

Oh, that's another $20 a month, right?

Yeah, I can see that, right? The glass cleaning robot's a good example. China already had— I saw them at the Consumer Electronics Show 9 years ago. They have glass cleaning robots that go up and down in your glass and clean it, but they have to be moved. Well, a humanoid robot could bring that robot to your house and put it on the glass and charge you, you know, oh, you need your glass— your glass is all dirty, man. I could clean your glass up up for $40, right? It'll take a day. Yeah.

And I'll just move the glass robot around the house until all the glass is cleaned. Yeah. I mean, I think that these things are going to, you know, I think the, the spectrum of monetization on these things is going to be pretty vast, um, of different ways that they can— these skills and these abilities can be monetized. I do wonder, do you see a robot ever becoming a content creator?

Yes. Your car will be— if there's a fire on the Golden Gate Bridge and your Tesla goes by the fire and it ingested the video off the cameras and made a Gaussian splatter, a neural radiance field, which is a new 3D scene. And it, you know, and it puts it on x.com into the fire feed, right? Isn't that— isn't that content that a robot created? Yeah, yeah, I could see a social network that's completely, completely done by robot.

Yeah, I can do that. I mean, even podcasts. I mean, that, that's a big discussion in the podcast industry right now is how far down this path path of content creation, um, do we want to accept in podcasting? Have you used Descript at all?

Oh yeah, oh yeah, I used— so Descript had me read to it for 30 minutes, and now I can type to it and it talks like me.

Yeah, yeah, I have a clone of my voice in there, right?

Bingo, right? So how much do you, as a podcaster, do you use it to fill in a sentence here, there where you made a mistake, or said, um, too many times or something like that, and you cut it out. And yeah, I can cut it out, right? And tell it to make you smarter, right? I said something stupid there, can you fix that please? And it could, right? Yeah, totally. You use that for 1 sentence out of 100, right? And then the next week to use it for 2 sentences out of 100, you know, another couple months, are you using it for 20 10 sentences out of 100, you know?

Yeah, I found that I used it too much early on, and then I've been kind of backing off from it a little bit because you have this kind of tension between authenticity and too much of a robotic type of consumption experience, right? If you cut it too much, then it— because it edits the video and the audio, I think the audio turns out better than the video does because true video gets a little chopped up, right? So you're kind of jerking around in the video while you're watching it because it's editing that, right? It's pulling out, you know, like maybe pauses, or it's pulling out spaces, or it's pulling out certain words, uh, and that causes kind of a disruptive kind of consumption experience. But really, my interest on that is, you know, voice cloning, the video capability, right? Because autonomous robots don't have to take the form of, uh, humans, right, for them to kind of take over, just like what you said. So there could be a whole podcast network that's generated based on AI technology or AGI type of technology that takes a subject and drills into it and creates very human-like experiences.

Yeah, and with less and less over time, um, human interaction or instruction, right? Right. It's gonna get better and better at writing or talking or doing a video avatar, kind of moving avatar, right? I can totally see AI content is coming. You know, it's still— it's hard to wrap your head around it because We don't think about what, what a robot could do better than, like, the Golden Gate Bridge fire, right? If there's a fire on the Golden Gate Bridge, I'm not there. For me to get there, it would take some time, but a Tesla is rolling across it every 3 seconds. Yeah, right. So it's already there. So a, a car can be like, oh, a truck just started on fire, right? Another one goes by, oh, it's burning good. Another one goes by. It's burning real good. And it keeps updating the Gaussian splat so people can watch it at home burning. And then it'll report when the fire truck gets there and it'll report how many lanes of traffic are closed and it'll report all sorts of things, right? And that's a news story. Fire on the Golden Gate Bridge and traffic is slow, right? Well, that could be written by an AI. There's no need for a human being to write that. There's no value in a human being writing that.

Other than, you know, yeah, I think it's already, it's already getting to a point where the video and the audio are almost indistinguishable from human experiences. We're already there. I can imagine what it's going to be, you know, a year or two years from now, and also what ChatGPT is going to look like when it comes out with 5.0 or 6.0 in the future. Or how much improvement is gonna happen in there. And it's hard to really grasp what the improvement could be in there. Is it just intelligence? Is that where the improvement is? Or is it just quality of output?

I'm not sure what that spectrum looks like. I mean, what makes a good— let's say I was comparing you, your podcast versus an AI podcast, right? Yeah. What makes you better? You're more knowledgeable at certain things. You're human. So you have a warmth to you, right? A little randomness. A little weirdness, right? We're biological and the computer is going to be cold. It's going to have a set pattern that it was programmed to do, right? And so the stories might not have as much detail as a human being right now.

Sorry, there's a leaf blower right outside my window. Oh, okay.

Of course, that always happens while you're recording. Of course, you know. Anyway, so, you know, over time, how do you improve the AI one to be better than the human one?

Detail, detail, detail. That much training data, right? It's the same thing that we're seeing with the, with the full self-driving stuff. It's, it's more, more training data in different situations and scenarios because, you know, I came across on my trip back from Washington, D.C. Scenarios that I'd never been in before on the freeways because of construction and those kind of things that it just couldn't figure out it because the bingo and we were—

it needs more data, right, to become more human-like, right, to have more knowledge about the world and more of a belief. On, on Twitter Space, a lot of times I'm talking to Dylan, who's a researcher at Unity Labs, and we talk about digital consciousness all the time. That's his favorite topic. Right. And so how does an AI get more human-like over time? Right. Detail, belief system, soul, a digital soul. I know a guy who's building digital souls for robots. Right.

So as the robot lives its life or operates, it becomes more human-like over time.

Yeah.

So we have to capture more data. From either humans or the environment or, you know, the physical world, right? Which, you know, just looking at the map that Tesla uses for their full self-driving is not complete, right? It's, it's missing. It's trying to infer based on the video cameras, but it's, it's still missing things, right?

Oh yeah. If there's a ladder in lane 1, it doesn't really know that well and certainly doesn't it doesn't know it the way it's going to know it in 5 years where like in Silicon Valley, there's so many Teslas here. There's often I'm in a pod of 4 to 8 Teslas, right? And so you start thinking about, oh, wait a second, the people up ahead of me could be imaging the road and oh, there's a ladder in lane 1. It could tell my car, hey, get out of lane 1 and there's a ladder up ahead. Right. And my car will just switch to lane 3 and I'll be like, Why did it just change lanes? Oh, there's a ladder in lane 1. It doesn't do that kind of stuff, but it will in the future as it gets more data, right?

And gets the ability to share. Yeah, I'm seeing that a little bit already with the software where it'll see something in the road and it'll kind of get out of the way, right? It'll just kind of like go around it.

But it's not doing it in a systematic way and it's not doing car-to-car communication like it's going to probably in the next 18 months. You know this is coming. Coming, right? Car ahead of you sees an object in lane 1, your car is going to switch its lane over to lane 3 because it knows there's an obstacle in lane 1.

Yes. So leftmost lane, right? So the most likely scenario is it's going to be Tesla to Tesla communication, right? So there's lots of situations where I'm driving down the road and there's a Tesla in front of me and one on the side of me or whatever, and it currently doesn't have any idea that they're Tesla's because it may not have a reason to, right? But when we have a reason to, um, then it'll probably put up on the screen that it's a Tesla, right?

There is some car-to-car communication because there's a red line telling you there's, there's traffic up ahead.

All right, so yeah, is that coming from cars up ahead of you reported to Google though?

Is that coming from Google though? No, it's coming from— the data is coming from Tesla's as well as Google actually is it from Verizon. It's sort of funny, right? I saw an accident on a two-lane road, and by the time I got up to the accident, the, the phone screen started turning red, right? Because people were backed up behind the accident and slowing down and stopping. And I was like— I, I asked a Google guy who works on Maps, I was like, how did it get red so fast? That, you know, because you didn't have very much data. There was, you know, 15 cars involved. Left. He's like, oh, we're buying the data from Verizon. I'm like, oh, you're getting real-time data from Verizon cell phones that it knows the Verizon cell phone is stopping on a road, so it turns the road red? Yeah, that's what we do. It's like, whoa, right? And people have no idea all these, or, you know, relationships between, you know, Verizon and Google doing real-time data sharing. We, we really don't think too much about that, but that, that's how these things work underneath.

Yeah. Yeah. So how would you recommend that just a person watching this, um, try and learn more about what the future looks like with this stuff? I know we've talked a lot about it here, but I think we've raised more questions than we've answered, probably.

Over on x.com, I went through 100,000 people's accounts and built lists for you. So I have 20,000 people in the AI world on 4 lists. I have a robot list just of the robot people. I have a neuroscience list just for the neuroscientists, right? So you watch those lists over time, you're going to learn just by watching and reading, right? That's one way. The other way is just dive into the industry and start, you know, do what I did. I, I made a database of 5,400 AI companies, right? So I could have some idea of what's happening in this industry and what I should be— what I care about, what I should pay attention to, and what's happening. And I can watch lists of all those on X Pro. And so in one screen, I can watch the entire world go by. If you're not using X Pro and lists, you're missing out. Um, but yeah, I built you the list. They're— they took thousands of hours to build, so you're not going to build better ones. I have them.

I have the lists that are bad by far the best lists in the tech industry. Yeah. So, Robert, you started your own show. It's called Unaligned. Yeah. Here, here in the last 6 months or so. And on your, your X account, you've got lots of posts that you put up there about all sorts of AI-type topics, VR, AR kind of topics. But I think you've really started to shift towards more AI.

Yeah, I'll come back. The AI-driven glasses or XR are mixed reality glasses, right? So I've been covering that world, spatial computing devices, right? I've been covering that world for a long time. And so as that gets hot again, I'll probably cover that quite heavily because that's highly interesting to me. All the AI I have, they're outside, but I have all sorts of pendants. And things that, new kinds of hardware that has AI inside of them. And I'm watching that world. I'm just really much more bullish on glasses because I think the AI needs to show us things, right?

In addition to talking to us. Yeah, and it's also a data gathering thing too, right? So it's seeing the real world from a human perspective.

Here's one, a fun one, right? You go to a national park, like we went to Antietam National Battlefield where the biggest battle of the Civil War Civil War was, right? I take a picture of the cannons on the field and it tells me about them. I take a picture of the church on the field, it tells me about the church. I take a picture of the signs on the field, it tells me about the signs. But, you know, because the signs just have a few words on them, it gives me a whole story about what happened at that mark, right? And then I ask it, can you simulate a battle on the field? And it can. It sends you a picture future. It's like, oh, this is gonna get really crazy soon, right? Because you can simulate all sorts of stuff and have it, you know, wait, what, could you show me what happened in history at this place?

You know, and it'll simulate it and send it back to you and it'll be somewhat accurate.

Yeah, yeah. I mean, it's, it's based on all the photos that they had, right?

It's a pretty freaky world that we're coming into, Robert, isn't of it.

Yeah, it sure is. Yeah.

And we haven't even covered music. There's a whole— there's so many topics to talk about. That's what I was talking about earlier. We could do a 2 or 3 hour episode here just on trying to cover the whole spectrum of this because it's going to impact every aspect of our lives. Absolutely. And, and, and it may, you know, let's spend a couple minutes talking about something. You know, we, we did kind of allude to kind of the exciting parts of this, but there's also some, you know, some scary parts too.

Technology always has a bad and a good, right? Cars— we all use cars because they have a good, but they kill people and they have a bad. They can kill people, right? Uh, 40,000 people in America die every year in them. So, right, so they kill people. Um, it's humans killing each other in the thing, but you know, it's It's, it's a problem. And so, yeah, every technology is like that. A database could be used to catalog, you know, people in a concentration camp, but it could also be used to find cancer, right? So it has a good and a bad, and humans have to figure out, you know, how do we do more of the good and less of the bad?

Yeah, and that's where this question of trust comes into too, is is that I think we haven't talked about Neuralink.

What happens to trust when you get a wire on your brain?

Well, does that turn you into an autonomous vehicle?

Yes, yes, it does. This is the problem becoming— that's one of the downsides of Neuralink is you don't have free will anymore. If, if there's a wire on my brain, I am not in control.

Somebody else is. That's a big problem.

Yeah, I mean, I mean, and somebody else is listening to all my inner dialogue, you know, which might be despicable at times. I don't want a government or an Elon Musk in my head, you know, particularly if I'm a young kid who's starting to figure out their sexuality, right, right, or something like that, where, you know, you don't want thought crime to happen.

That's dystopian. Yeah, so, so we got a comment from a viewer of this, is that this does feel like listening in on a conversation between two people on a park bench, Robert. That's the way we are, right?

Or in a Starbucks or something like that, right? I remember driving around with you in my car in Seattle, right, when we worked at my Makes sense.

Exactly. And I think you recorded that as a video back then too. That was before people were really doing that kind of stuff. Yeah. So, and then you came out here last year. Yeah. And we met, we caught up with each other, and we both had Teslas with us at that time. But yeah, it's kind of a strange world in And I think a lot of people feel uncomfortable with what's developing and the way it's moving.

It's a freaky time. A humanoid robot with cameras, you know, you're going to have it in your house someday. Maybe it takes 10 years, but you're going to have a robot like that. That's freaky, right? What do you mean? I've had these arguments with family members about, you know, Amazon Alexa device that's listening to you, right? Oh, wait until you got a robot that can move things around your house and see everything and catalog everything and track everything in your house.

Like, it'll know where your ketchup bottle is. Oh yeah. Well, also, I don't know if you've had this happen to you too with those, those, you know, in-home kind of audio devices. I don't want to say their name, but, um, because you got a lot of them that'll listen to you, right? Disrupt our show. But hey, Amazon. Yeah, exactly. So these, you know, and I think they're going to convert into almost like AI robots as well. It's going to be audio, but it'll have the ability to have reach. I mean, I think most people have the ability to control their home, their lights, their, you know, built-in functions of their home, right, with a voice command.

Can't see. I have digital lights with microphones and a computer underneath, right? And you can change those, the color on those on your phone.

And, and there's microphones so you can do, uh, like EDM concert kind of music. But I do kind of wonder, kind of as we move into this, um, you know, are we going to be able to trust any of this stuff? And I think as wise for us to, to be skeptical of the darker side of this kind of stuff. Yes, it's also, you know, it's coming, right? Right.

And we either have to accept it or live in a life of anxiety, right? Or go Amish. Or you join the Amish. There's 30,000 people who don't have computers, and, you know, maybe you have a little phone, but you know.

Yeah, that's a whole nother level of cut the cord, you know.

I think it was— yeah, you know, you can go join them, you can start your own cult. We don't allow those newfangled AI computers in our cult, you know.

Right. And I do think that there's going to be a certain amount of people that are going to do that, um, because of the, of the fear factor.

What do you mean a robot's going to come in my house and watch me naked?

No way, not in my house. Awesome. Well, it's also going to potentially enable, um, better law enforcement too, right?

More, more, more control. A guy broke my, my, uh, uh, window in my Tesla out front and destroyed a bunch of other stuff, right? Was swinging an axe around.

Um, the first thing the cop wanted was the, the door, the doorbell footage, right? Right.

Or the Tesla footage. I mean, yeah, he asked for all the footage. And then of course I had no footage because I had— I turned off my Tesla because it was out in front of my house. I figured out why do I need the surveillance of myself? Well, now it's on. Yeah, because of this incident, right? And I've made sure that the, uh, Ring— there's a Ring device on my garage that looks at the whole driveway, right? And so that's on now. And then I have a Ring doorbell that that's taking video of the front of my house too, and that's on now. So next time the cop needs some footage, I'm gonna have some footage.

Yeah, totally, totally. And, and to some degree, that's, that's a little bit of a deterrent, right, from, yeah, from criminals, you know, attacking your home or something like that, is that they, they know they're going to be videotaped. If you— I mean, I've got, I've got 3 4— no, I've got 4 Ring cameras myself that are pointed around the house. So, so if anybody tries to approach my house, they're going to get videotaped.

So, so, uh, one of your readers— I see it on the screen— just said, uh, $16,000, wash the dishes, and you get an error. You know, yeah, I have a feeling it's not going to give you an error on anything you ask it to do because it— the Stanford robots know how to do thousands of tasks. Right? Yeah, it's not just wash dishes. It's like know how to do anything a human being wants to ask it. Yeah, I mean, go garden and weed the garden, you know, bring me a Diet Coke, wash my clothes, iron my clothes, clean the toilet, right?

Yeah, I think what we're seeing with that comment from, uh, Smirk Station is, is a skepticism of the, of the you know, reliability of the tech, right? That, you know, and we've seen this even with Tesla, um, that these things don't start out being perfect, right? They do have errors and flaws and things that it doesn't yet know about.

But it is janky, right? I saw a robot clean up a mess on a table and it was janky.

It wasn't real smooth about it or confident about it like a human would be, right? But And over time it gets better. It's the same thing with the Tesla Full Self-Driving. I'm waiting for the next update to the Full Self-Driving stack that takes me into the, the 5.0, um, spectrum, the 12.5, you know, XX or whatever that's coming. And, and I'm— this is another factor too. So you're going to have a certain amount of, um, obsolescence that is built into this too, where where, you know, in my Model 3 and your Model 3, because I know you've got a 2018 one, is probably on hardware 3, right? And that may not be able to go much further because of the processing capability that is needed to run these more advanced neural nets. Yeah, I do wonder, I know Elon's working really hard to try and get the next version of full self-driving onto to, um, the Hardware 3, uh, cars because there's a lot of Hardware 3 cars out there. I think there's over 2 million, right?

Um, yeah, so there's definitely a lot. Those are likely to the cars that we would put in a Robotaxi network, right? If you— I just got my wife a new Model Y, well, maybe the Model 3 gets in the Robotaxi and goes, makes us some money, right? And does some work.

And I don't care that that gets trashed because it's an old car anyways, right? Yeah, and some people paid, uh, $15,000 for Full Self-Driving on a Hardware 3 platform. Yeah, so, so, you know, I think Elon may be coming into a pickle here because, you know, the cars are not capable of being upgraded to Hardware 4 or coming Hardware 5 capabilities unless they somehow figure it out. Yeah, I don't know that they're going to be able to.

Um, it's gonna know pretty quickly what it can't Yeah, like a Model 3. What, what can it not do that a Model 4, uh, Hardware 4, which has I think 3 times more compute on it, right? Um, and can run a bigger parameter large language model and have more accuracy, right? Okay, it could be smoother, right? It could be more luxury, more, more refined of an experience. It's more human-like bike, right? That's what the bigger model gets you in writing too, right? If, if you use ChatGPT, the big model, you get a better answer than if you use the small model, right?

Well, and we still don't have AI in the car yet, right?

Uh, as far as— not like that, right? But you know that's coming as well, that you need a large language model in the car with like a Grok in the car that you can talk to. Hey Grok, where's, where's a cool nightclub to go to tonight in Las Vegas?

We're going to Vegas Yeah, because a small thing like, you know, like I was on an 8-hour drive with my Model 3 last night, and even a small thing like asking it to stop at a gas station or to find a bathroom or something like that, it didn't quite know how to handle that question.

And it has the data, by the way. Certainly, if you look at Superchargers, right, it tells you to stop at, you know, Harris Ranch Supercharger. Yeah, yeah. And I did this right there. If there's a bathroom nearby or a restaurant or a hotel, right, right. So it knows, it, it has that in the database.

It's just not hooked up. Yeah. Eventually I can see the car, you know, learning about my patterns around things like that and prompting me, you know, do you want to pull over or something like that? Are you needing to stop? Are you— I mean, I, I can get kind of beginning to go at a deeper level of managing the, the process of going on a long drive, right?

Back to trust. What if you ask Grok, you know, take me to an In-N-Out? Well, there's 3 In-N-Outs near me. It knows— it could know the line length at each In-N-Out, right? There's Teslas in line. It knows how long the line is. It has video of that, and it could count all the cars in line, And certainly as soon as a Tesla goes through the drive-thru, it knows how long it was in the drive-thru. So it knows the average wait time at that specific drive-thru. So your car could know, oh, this one has a short line. Let's take you there. Yeah. And that's trust because maybe you don't like that in it, right? But you're being driven there because it has a short line and you have to trust that the computer made the right choice for it. You. And that's a lot of trust to give a piece of melted sand, right?

It's going to have to be proven that, uh, it, it delivers on the promise of the future, right?

And yeah, it takes you the right in and out, right? If it takes you to a closed in and out, you're not going to trust it, right? Yeah, yeah. You're gonna go, this is stupid, it took me to a bad place, it didn't give me the right thing, right? And, and you lower your trust, and so now you don't believe it the next time you ask for an In-N-Out, right?

Yeah.

But I have a feeling it's going to be way better than any of us, and therefore we're going to learn to trust. I already trust it when it takes me off the freeway. It's like, why are you taking me off the freeway? Well, I see it, there's a bunch of traffic on the freeway. Oh, that's— you took me off the freeway to take me around all the traffic.

Oh, that's smart. Yeah, it does that already.

I think it does that already. It does. What— one day I was taking the picking the kids up, and it took me off my normal, you know, pick the kids up from school route, which is usually the same streets. Took me on a different street. I'm like, what? What? What? I, in fact, I overrode it because I didn't trust it. I overrode it because I was like, no, it'd take me to my usual route. It turned out there was a bunch of cops closed the road. They knew about that, and it was trying to route me around it. Yeah, I've had to— after a few of those kinds of situations, you learn, oh no, we better go with what what the computer's saying. The computer's better than us.

It has more data about what's going on. Yeah, it does get back to that to some degree. I, I can totally, totally sympathize with that thought because of the navigational routes that it takes. I mean, sometimes the car takes off and goes on a, on a route that I wouldn't even have thought to, to do, and, and I just have to sit back and trust it and say, well, well, maybe this is a faster route that it was able to detect that I would have no idea, uh, would be a possibility, right? On the side streets or something like that. I can just imagine trying to drive through LA or something like that. It would know all the back streets to avoid the main freeway, right?

So that's what I— one thing I want Grok to do. Hey Grok, take me the scenic route, or take me the fastest route, or take me the smoothest route, or take me the cheapest route, or take me the safest out, right? My friend who runs an insurance company has a map of every street in America, how safe the street is. You know, I don't know— I didn't know that until I, uh, had a friend who worked in the insurance business. But, but every street is different safety rating. Yeah, it has a different accident level and a different, you know, consequence for driving down the street. And so it could take you the safe route, right? It knows which route is the safest. It just doesn't present that yet. Right.

But an LLM could go through all the various routes and bring them to you. Well, they could probably map all of the— in the training data, it could map all the locations of fatal accidents or something like that. And, you know, put a weighted factor into that route and its risk factor. Now, granted, the amount of crashes in autonomous vehicles like a Tesla are dramatically lower than human-driven cars already, right? So I kind of wonder about that to some degree. It's like the insurance companies are not factoring in whether or not a Tesla has Full Self-Driving in it. And I'm a little torn whether or not I wanna tell my insurance company that I'm using it at this point. But at some point, I would think that would be a pathway to hit lower insurance rates.

Uh, all right, so a human and an autonomous car might start at the same safety rating, right? Uh, and safety rating is, does your car hit somebody else? Because that's when an insurance claim comes to you, right? Right. If somebody hits you from the back, they're paying. So it causes you problems, so you don't want to have that happen, but that person's paying, not your insurance company. Company. And so, uh, um, over time, the accident rate in an autonomous car is going to keep going down. The human car is going to stay the same or even go up.

Humans are actually getting worse at driving, right?

Because of full self-driving. Yeah, because we're all staring at our cell phone. Oh, boom, we hit somebody in front of— oh, shit, you know, right? Um, and an autonomous car won't do that. My car never hits somebody in front of it, right? If somebody slams on its brakes in front of me, I'm slamming— my, my car slams on its brakes then stops in time, not a problem. The person behind me who's on their phone may not be— may not be— slams into me, but they're paying the insurance, right? So the autonomous car, the insurance rate keeps going down over time. The wholesale cost of the insurance keeps going down over time. So if the price stays the same as a human insurance company, then the profit is going to increase over time. But if, if they reduce the price, then the price will go down over time. And I would assume an insurance company is going to reduce the price because they're competitive and they want, you know, hey, Tesla owners, come and sign up on us. Or Tesla's going to do its own insurance, which it is, right? Yeah, yeah, yeah. It's always— Elon knows this. He, he knows that if I charge somebody $100 a month for insurance right now and they're using Full Self-Driving 20% of the time, As they increase their use of full self-driving, because it's getting better, right? Next year they'll use it 40%, and the year after that they'll use it 80%, right? Because it get— it's going to get really, really much better over the next 2 years. So the insurance rates are going to come down for like an e— for a Tesla, and they could reduce the price and really eviscerate the rest of the insurance industry because The rest of the insurance industry has to support gas, gas cars, cars.

Yeah. And Elon's been saying for years that this robotic technology, the autonomous stuff, is going to change the economy. It's going to make a, you know, the cost of labor basically almost go to zero. And I think to some degree that that's a reflection of what you're saying there is that—

not zero, but— a lot less, right? Because my Tesla sitting in the garage right now, it costs about $2 an hour for financing and everything. And well, you put a human in it, that goes up to $20 an hour, right? Right. And so eventually an autonomous car is going to be able to beat an Uber-style, like human-driven Uber-style car by a lot and have a better experience because The autonomous car is going to be consistent and it's not going to talk too much. It's never going to be drunk or stoned or anything like that, or never going to be tired because they've worked all night long, right? And it's not going to smell because humans smell once in a while, right? It's a better experience. It's a more consistent experience. It's a safer experience, right? So autonomous cars have a lot of advantages there, and, and they're going to be cheaper.

Someday. Yeah, so I also heard some speculation that the new Robotaxi that's going to be announced here in a month or two, um, is going to have an automatic inside cleaning capability. Because if you're going to be in a Robotaxi, right, there's going to be, you know, sanitization that needs to happen in the vehicle, right, between riders. So it's, it's an interesting aspect of it too.

And that also like a built-in little vacuum robot that'll just go through and clean up everything and validate it and go, okay, this, this one's ready to go, there's nothing wrong with it, right?

Or have it come up through the seats or something like that, and then it quick dries or something like that to, to sanitize.

It'll be interesting to see what they do there. Yeah, yeah, no, that, that makes sense. And, you know, have a self-driving, self-cleaning or self-validation, right? Oh, this one's clean enough and there's nothing wrong, you know, right? There's nobody cut the seats with a knife or anything like that, right?

Or barfed in the car and left a mess. Well, that's always the risk, right? Because people that maybe drink too much are more likely to take a—

yeah, like a taxi or an Uber taxi. Or you have kids, they get carsick, right? You know, I've had kids barf in my Tesla. By the way, they clean up pretty good.

Even the white seats, people's like, oh yeah, the white seats, they wipe up real good. Yeah, yeah. So it's gonna have to be some sort of automation on that because I mean, it's, it's not like you're going to take it to a detail shop after every ride. No, and it needs to validate and, you know, and then the concern about, you know, viruses or something sharing and things like that.

I'm sure there's sensitivity to that too, right? Right. Um, there shouldn't be too much of a viral problem as long as there's a few minutes between riders, right? You know, maybe if somebody sneezed on the steering— well, you shouldn't be touching the steering wheel. Why are you touching anything in this car? Do they leave a cloud of virus in your, in your cabin? Well, that gets cleaned out in a few seconds, right? So it does. Yeah. Yeah, so it's, it's a new world. But you're right about the autonomous cleaning. I, I think that's a big part of doing it right, because the system has to validate that car is good to go and good to go back into the, into the fleet and pick somebody up, because it needs to be consistent. Consistency builds brands. I talked to Sebastian Thrun, who runs Waymo, who ran Waymo, started Waymo, started Google's autonomous car industry. Industry, uh, company. And he, he was like, we've talked about what, what makes a great brand like Starbucks. Why is Starbucks so good? Starbucks is actually shitty coffee. It's not, it's not the coffee that sells it, it's the consistency, right? Because humans can get used to shitty— I'm drinking some Starbucks right now, right? It's shitty coffee, but I got addicted to it because Because human mind, I learned this, the human mind can get used to shitty things, right? And actually starts liking it because every morning you get the same shitty coffee, you get used to it, you get, you know, you get the caffeine out of it and whatever else you put in it. The Starbucks, what makes Starbucks really special is the consistency. You get the same latte in London and Japan, Seattle. Right? It's exactly the same. The experience is exactly the same. And ubiquity— there's 3 of them in one mall in Yokohama, Japan, right? So you're in a mall, you don't even have to walk very far to get your Starbucks. Yeah, it's like, where are you in the mall? There's 3 of them. It's like, right? And so Uber has ubiquity, but they don't have consistency, right? Oh, it is wildly inconsistent. Right, right, because the car is inconsistent. Sometimes it's a new Mercedes-Benz that picks you up, sometimes it's a shitty old beat-up Toyota, right? Yeah, right, I've gotten in many shit beat-up cars before, right? Yeah, right. And so as Tesla, you know what you're gonna get. Even if it's beat up, it's a Tesla, it's safe, and it's, uh, you know, fairly fun to be in, right? Because the batteries are underneath you and it It's fairly good at driving around. It's actually a really great car to drive. And two, if you get rid of the humans, the humans, the inconsistency, you get rid of the humans. Now you have a much more consistent service that's going to make the brand resonance around such service go up. Right. And so in 10 years, when you get into Vegas and you need a ride to take you to the hotel or to the convention center or something like that, got. Which one do you choose to use? Do you, do you order a Waymo or a Zoox or a GM Cruise, which is Uber, uh, or do you get a Tesla? Yeah, well, the Tesla has a better experience inside, so I know how this game is gonna go down. And the Tesla is built with less technology, so it's cheaper.

Yeah, well, I think that's the key advantage that Tesla has, the ability to make things that are are mostly reliable, right?

And for cheap— cheaper, right? Can't— LiDARs cost a bunch more money, and, and then you have to have two computers to process LiDAR and camera, because like the Waymo camera car has like 19 cameras on it.

Tesla only has 9, and it has a whole bunch of LiDARs. And which— so where's the, where's the 9th camera on the Tesla?

Is that a new addition to the— there's 3 hardware 4s. 3 in the front windshield. There's 2 more, that's 5. There's 2 more, that's 6, 7. There's one in the back, that's 8.

And there's one inside, that's 9. Oh, okay, got it. Yeah, I think I forgot to mention the one that's in the rearview mirror.

Yeah, yeah, it's always— the machine's watching you.

Oh yeah, it knows how many times you pick your nose, you know.

And it's not always able to—

it doesn't do anything anything like that, but it could. Yeah, I was driving in the dark last night and occasionally I could tell that it couldn't really see my eyes because, because of the, the light in the cabin wasn't, wasn't that bright. So, so, oh no, it was flashing on the screen and I'm like going, you know, I'm shaking the steering wheel and adjusting the speed and stuff and it wasn't recognizing. So it was definitely prioritizing its vision over even touching the steering steering wheel.

So yeah, yeah, right now it's— for the next year or so, it's going to be in a dangerous state where it's making a lot less mistakes than it used to, and therefore humans are going to change their behavior of using the vehicle and start sleeping in the vehicle or looking at their phone more often or something like that, right? And then it makes a stupid-ass mistake and kills you, and that's the problem. So they need the human still to be supervising the machine right, to, to keep the safety rating up. Yeah, eventually, you know, 18 months, 36 months from now, it's going to be so good it's never going to kill you. So that's when they can turn off all the nags.

But yeah, it's the— a few years away. Unsupervised, not, not the current supervised, is what— yeah, is what it's going to take, probably, for the—

I would think for the human taxi. Yeah, drive around without a human in it, right? Uh, Waymo is already humanless in San Francisco. You, right?

No human needed. That's pretty cool.

Yeah, but it also is a limited capable area that it can operate in too, right? Yeah, but actually talking to them, they'd say no, that's sort of just, uh, uh, it's temporary. They told me it's temporary. Oh, they can drive everywhere in America. They just choose not to because they don't have the fleet to support that, or the business model, or the new workers, or any of that stuff, right? So they're starting in just a few cities like Phoenix and San Francisco, getting it good, then they're going to grow it to more cities. How many years before it has half of all of American cities? I think there's 300 cities with 100,000 people or more in America, right? How long will it take to get half of those? And then how long will it take to get all of those?

Yeah, not many years. Not many years. So Robert, kind of the, the last topic here, and we will call, call this, this episode a wrap, but, but I wanted to ask you, what do you think, um, the skills are that humans need to gain right now to prepare for this shift? I mean, are we— is it kind of, kind of trying to adopt this technology like you have to learn about it so you're not, you know, getting behind? Is that the key? Or, I mean, what kind of skills do you think people need right now?

That's one. First of all, understanding what it does for you and what it doesn't do for you and how that's changing. Because a UI AI today that doesn't do what you want it to, right, if you're asking for a business plan or something, and maybe it doesn't do a good job is that going to be true in a year? Is it going to be true in 2 years? Is it going to be true in 3 years? In 3 years, no, it's going to be really good. So you have to, you have to get involved early. The earlier you get involved, the more you'll understand how to control the machine, how to talk to it, how to get it to do things, right? Um, and you'll also have a much better idea that your job's going away, you know, if it is. Right? Because if you're watching the 5,400 AI companies that I have on X, you can see it coming. You can go through them all and go, oh, that company could do away with my job, or that company could, right? Um, and so understanding that's important. The last one to get fired is a, is a person who knows how to use AI because they're going to be more productive, right? If you're, if you're a programmer and you're using AI, you're twice as productive as a human who doesn't use AI already today. So if you're not using AI as a programmer, you're an idiot.

Yeah, um, you're gonna get fired.

There's also— you're not gonna have a job. You're gonna be the first to get fired because all the rest of the team is using AI.

They're twice as productive as you are. Well, that, and also, is it even secure with that?

Isn't AI going to be able to write code? Oh, it does, it does. But right now you need a human to tell the thing what what to write.

You know, I need a 20-year-old or whatever, right?

Prompt it, right, to be able to— yeah, you need to tell it what, what, and you need to fix it too, right? Because it bullshits once in a while. You gotta— if you're a computer scientist, you know how to fix it. But the people who are using it at an advanced level are way ahead of the other programmers who aren't using it yet, right?

Yeah. And I think a good example of, uh, even AI makes mistakes is, uh, GPT-4 Omni is creating images, it constantly makes typos in its images. So if you want it to output text, and this is maybe an indication that we're, we're at a phase where it can't really create error-free code, right?

If it can't create error-free images with text when it's an LLM, it's getting better. So how many more months will it be before it fixes the pro—

right, you know, and I've been waiting for like a year, but—

and there's a whole bunch of companies who are trying to fix that problem. I just interviewed one yesterday. Yeah, who's— they're building a system for enterprises that, uh, learns iteratively and refines the, the business model that you're building so it gets better and better and better and reduces hallucinations over time, right?

Yeah, I've also seen, uh, these AI models battle with each other too. Like, I'll create something in ChatGPT-4 Omni, and I will run it through Grammarly, which is another AI engine, and it'll tear apart what I learned yesterday.

Yeah, that— okay, so there's one technique: you can spit an answer out of a large language model like ChatGPT-4.0, right?

Yeah.

And then bring it back into the model and have it rate itself, have it rate its own answer for accuracy and stuff. Like that, and it actually does a pretty good job. But if you give the same answer to, let's say, Llama from Meta Corporation, a different large language model, much more accurate.

Yeah, and I talked to a grammar expert on my podcast tip show that I do, and I told her about this battle between ChatGPT writing and Grammarly, and she told me that she prefers ChatGPT that GPT's writing is more grammar, grammar, uh, accurate than Grammarly.

Yeah, I thought was interesting feedback on those models. Well, the large language model that OpenAI has has more parameters than the Grammarly model, right? And it has a lot more engineering behind it because they, they got $10 billion and they have a lot of NVIDIA cards. It's an NVIDIA— if you want to know who has the best models, look at who has the NVIDIA cards because they're probably the ones that has the best models, right? I mean, Tesla's model, uh, XAI's model Grok just got way better in the last week. Why? Because he got in new NVIDIA cards turned on, and all of a sudden his models are getting better because he can give it more data and, and choose faster, train it faster, right? And better. And train it better because you have— if you have 100,000 NVIDIA cards and you're using the same data set, it can give you a better parameter model. It can make a finer model, right? Because it has more compute to go through it more and more and more times to build that model.

So Robert, I know you're always looking at this stuff. Is there anything that we haven't talked about that you think are revolutionary and what we can expect here?

Uh, I was just talking to a couple investors, big investors, and they're talking about the same kind of things, you know, virtual beings are coming. You know, and they're going to be pretty crazy because once you get a pair of glasses on, you could have, you know, something that looks like a humanoid robot standing next to you, but it's just all virtual, right? A virtual being, and it can talk to you and interact with you, high-five you even, right? You know, shoot it right in a gun in a game. Yeah, you're non-playable characters are going to get pretty smart pretty soon. Yeah, you're going to have conversations with them before you shoot them.

Yeah, and you're going to look completely crazy because you're going to be looking like this and talking to nobody, right?

And then you're gonna shoot it in the head with it, you know, like, I don't like your answer, I'm gonna shoot you.

That's where you have all these people walking around in the world talking to, uh, the virtual agents that nobody else can see.

Wait until there's a new kind of paintball game or something in your neighborhood. You're wearing the glasses and all sorts of mayhem is happening virtually out in the street. You got to go and, you know, shoot some things and get some exercise. Yeah, yeah, that's coming. I, I, you know, the guy who started Metaio showed me monsters on the sides of skyscrapers like 13 years ago, and he got hired by Apple to build the strategy for the Vision Pro, right? So I still don't have that. I'm still waiting for monsters. He promised me monsters. I still don't have monsters on the sides of skyscrapers, really. So I I was talking to—

It's close with the Vision Pro. It's close, but yeah. Yeah, so I was talking to a guy from Apple on my trip to Washington, D.C. this past week, and he was telling me that there's big things coming from Apple around AI. Yeah. Is there any glimpse that you can give us on what that might be?

Well, they have Apple Intelligence. All right, so they have a new system, a new series coming. Whether it's really amazing or not, right now, nobody thinks it's really amazing. Amazing. But it— there's a new AI coming, and he's right, they're spending a lot of effort. I mean, I, I see one of the guys right here, you know, who's, uh, building infrastructure, MLX infrastructure for Apple, you know, and they're doing a pretty interesting set of things. Um, the, the other thing, like this M1 processor I'm talking to you on, you know, because I'm talking to you on an iMac, 21% of that processor is for neural— is for AI inferencing. Most people don't haven't even used that part of the chip yet. I, I have 4 of them in my house. I don't use it. And I'm an AI influencer who supposedly uses a lot of AI, you know. To build a summary of a transcript only takes a few hundred milliseconds, then it's back to not doing anything. Oh, interesting. So there's not enough for it to do yet. Well, you know, by the end of next year, I have a feeling there will be a lot more things for that chip to Do you know if the Intel i9 processor has anything in it for AI? They're behind Apple and others. They're trying. Intel just added— a lot of the PCs are getting AI capabilities, but other companies are ahead in terms of trillions of operations per second, which is what, you know, what it's going to make the robot of the future or the virtual being of the future do a lot, right? So, or the holodeck. We're getting a holodeck soon, right? Because right now I'm looking at you on pixels on a small screen. Well, what happens when it's volumetric pixels and there is no screen? It's— you're in my room, right? That's right.

That's what, that's what's coming is a completely virtualized environment, a digital twin around you. That's a 360-degree recreation, right?

Is that what you're saying? Whatever you want, right? It's all voxels, right? You go to the Apple Store and put on the Vision Pro, right? And you see the Apple Store. You're not seeing the Apple Store, you're seeing a computer screen, right? Everything is created for you. Yeah. So you— and you start going, oh, wait a second, Uh, every voxel can be changed by the AI computer, right? So I can change the Apple Store into Yosemite, and it does. You turn the knob and you're in Yosemite. You turn the knob back, you're in the Apple Store, right? And soon you're going to have that in your house with more capabilities, right? Freak, freak, and more freak. Midjourney is building a holodeck. Unity Labs is building a holodeck. Yeah, and if that— if those two are building holodecks, I assume a lot of other companies are building holodecks because that's how you're going to compute soon.

So I think we've already invented the replicator, right, from Star Trek, right? Yeah, the holodeck. Yeah. So is there anything missing off of the—

off of the Star Trek, uh, tech playbook that we haven't recreated? The transporter. That's going to be a hard one. Now, I could transport you digitally into your robot, right? So if you're in Germany on a business trip and you have a humanoid robot in your kitchen, you could jump into your robot and now you're really in your kitchen, right?

Moving things around, right?

Yeah, you don't have to take the risk of being de-atomized in the process, right? Bingo. We can do this digitally but not Not physically, right? Not, not with our bodies, right? You can't cut that up and translate that over the internet. Yeah, but we can embody someday, right? We— I— if you were wearing glasses, I could embody myself into your head, right, and experience life like how you're experiencing it. So if you're in the front row of a Taylor Swift concert, share them for that. Yeah, share your glasses with me, man. Come on. Don't hold out Taylor Swift on me, you know. Yeah, right. Maybe even charge people. Yeah, you want to watch Taylor Swift in the front row? You know, I spent $5,000 on these tickets. I need some money back.

$20 for each of you, you know, or $50. That's what I was going to say.

You could sell access, right? It's— yeah, right. Bingo.

Reality TV, you know, a new kind of reality TV. And then Taylor Swift could take a, take a rev share on that.

Well, now, are you, are you embodied into Taylor Swift's glasses?

Ah, there you go. Now that would be a probably a more valuable connection, right?

And, and, you know, then it gets weird because once you do that, then you have professional video people around, but, and you can pop from person to person. If there's a front row at the Taylor Swift concert and everybody is wearing Apple Glasses or Apple, some are wearing Apple, some are wearing Meta Glasses, you can embody yourself in any of those people, right? And the computer can join all of those cameras together into an array camera, a light field camera, because that's what Stanford's first light field camera worked. They had a grid of cameras, right? Right? And so now you have people standing next to each other wearing cameras.

You could join all those together and make a 3D scene of the Taylor Swift stage, right? Yeah, I think you've come up with a new, uh, way for people to, uh, have a job, Robert. Bingo!

I'm signing up for that job. You mean I got to go to concerts every Friday? Okay, sure, that sounds fun. Yeah, exactly. Make a couple grand every, every weekend because I'm going to some cool concert.

Yeah, yeah. Like we've seen in the past with technology, it's, it's created, uh, income opportunities in ways that we couldn't have predicted necessarily, right?

Now, the truth is, I, I probably, if I was that kind of entrepreneur, I would be creating a virtual world with a virtual stage, and I would be creating, you know, my own music on that, or my own experiences for people, right? And selling that. But yeah, people are gonna pay for all sorts of weird shit, you know. I'd like to watch a guy run a sewage treatment plant today. That's $20. I don't know why you want to watch that guy work, but you know, it's sort of interesting for an hour, you know.

I think we've seen some of this stuff happen before, like, um, um, like back in the early days of the internet. Dirty jobs, new kind of dirty jobs. Like personal vloggers, right?

That would mount a camera on their head, right? Yeah.

Justin Kan, who started Twitch, before he started Twitch, he—

iJustine is another one that did that. Bingo, right? But Justin had 5 cell phones in his backpack all multiplexed together so he could always have a signal.

And I think it was a live cam is what they called it back then, right?

I think it was a live cam. Life cam, yeah. Bell, Gordon Bell at Microsoft Research was wearing a life cam and did a whole bunch of research of wearing a camera permanently.

And really. Well, you tried to do it with the Google glasses for a while too.

I did, but the Google Glass, the battery didn't last very long, you know, so if you're like. If you're streaming or recording on it, it could only record for 45 minutes before the thing died. That's what a lot of people like— are you recording me? I'm like, fuck no, it takes too much battery to record you and you're not interesting enough to record yet, you know?

Yeah. And if people knew that you were— you had a video camera on there, they'd be less likely to come up and talk to you, right?

No, no, that was— I didn't stop them.

No, in fact, that encourages—

it encouraged them to do that. That occurred At Coachella, Thomas Hawk and I had big 600mm lenses at this music festival. And just having those lenses in the field, people would come up and dance for us. They wanted to be photographed, right? There's a whole contingent of human beings who see a camera and they start performing for it. Right. It's just innately, you know, it's like, oh, there's a camera.

Let's start doing something goofy.

Yeah, right. Create a TikTok. Bingo. It's infinite. I mean, this is the new job though, right? As we kill old jobs like driving a truck or something like that, there's all sorts of new jobs coming on the meta in the future of this stuff, right? Because we're going to be walking around and want to play a new game. You're going to have a shack with new games and you're going to sell me new games. It's— or new clothes or new guns or new, you know, experiences or new concerts or, right, or new, new kinds of art museums or new kinds of nightclubs.

They're all coming, right? It's all really based on the attention economy, which has been around for a long time, right? Yeah. I remember we talked about that a lot in the early days of the internet was how people are gonna have full-time jobs around capturing and maintaining attention.

Yeah. So, and that's what's happened. At Microsoft, we had a Second Life island back in 2005, right? You know, and the only reason that didn't work out is the technology was too early. You could only have 100 people on an island. And we paid $100 grand to run the island, you know, to build an island. And I remember that if I could get a few thousand people on that island, it'd still be operating today, right? Right. It's a great way to interact with people who are your customers. It's much better than going to Google Meet or whatever, right?

Yeah. Or Zoom. Well, Robert, it's been fun living in the future and and talking about what's to come and how it's— how to maybe trust some of this stuff just based on us having better knowledge about what's coming.

So run your models locally, right? Well, I think that if you don't trust it, if you don't trust the man, don't give them your data. Create your own LLM right on your— well, or use a lot— use an LLM that you can run locally. On your own machine and not be connected to the internet, not give, not give the, you know, the cloud computing any info about yourself. But you know how humans work.

If it's more convenient to give the man the data, you're going to give the man the data.

And, and you're the one that's being sold many times. If you look inside a large language model, they've read my blog for a decade, right? And it knows a lot about me.

It really does. It's scary.

And it's probably growing in its knowledge. That's probably why I'm willing to trust it because I'm already deep inside of it. Right. It knows— you ask Grok, show me all points of view about Robert Scoble. It tells you what I've done bad in the world and what I've done good. Fit in the world.

Yeah, it's a new, new privacy problem.

Everybody knows your shit. You better be careful. It's like Santa Claus, he knows when you've been naughty or nice, you know, and he's keeping a list, you know.

It's—

and there you go, all the time, right? All the time.

Coca-Cola is going to have all your naughties. Well, Robert, let's talk about how listeners can get a hold of you.

And you got it on the screen, Scobalizer, right?

If you can't find me, you can't find anybody.

So, right. Oh, that's, that's a good point.

Put my last name, you'll find me.

And then it's your YouTube channel that has your, your videos that you have up there. Yeah, that's one of them.

The, the Unaligned is on another one, but you can figure it out from x.com. Sure. Yep.

Well, thank you so much, Robert, for spending, what, close to 2.5 hours with me.

Yeah, it's fun. You're awesome. Well, thank you so much. I appreciate you taking time. This is actually the first time I've—

well, it's been a few months since I did the last Trust Factor episode, so, so— well, we covered a lot of things.

There's a lot of trust problems coming, so— right. And I'm, I'm just starting to unpack this. This is the 13th episode of the series, so I've been diving into this, and a lot of the topics have been about AI.

So, so I, I plan on digging in deeper to other aspects of trust too that are— we didn't even talk about intellectual property. I mean, if you're a business and you have a whole bunch of plans for your next product, do you want to give that to ChatGPT?

I would think twice, right?

Right, that's true, that's true, because you're not getting paid for that. Are you going through the API door, or are you going through the front door. Because if you go through the front door, they use your data to train models. If you go through the API door, like you're a software engineer, they claim they don't use that data to train. So even right there, you have to know the difference, and you have to know what you're doing, which most people don't.

Yeah, I think— isn't OpenAI under a fair amount of lawsuits right now for copyright infringement?

Uh, well, that's because of how they train the models, right? They just ingested everything. They ingest— that's why it knows me. It ingested my blog for 10 years and stole it to make the model, right? Yeah, they didn't pay me. I didn't get a paycheck in the, in the mail for this, right?

Yeah, but they just— I think you'd have to— yeah, I think you'd have to comb through the terms of service with YouTube too to find out.

I'm sure that they're using all those videos as training data. And it changes. Uh, uh, x.ai just changed, uh, you know, uh, Twitter's terms of service so that they can use all of your data unless you change it back to say no, you can't, right? And most people don't even know that, right? So most people haven't turned it off. I actually leave it on because I want the AI to have all my data because it already has all my data anyway. But the more data I give it, the better it gets. Right. And this is the problem. Everybody's going to do this. And, you know, people get wound up. Oh, no, I'm not going to do that. I'm like, yes, you are, because you're going to have a robot in your house or at least, at least delivering pizza to your front door.

And it's going to look inside your house. Go. Nice house. Yeah, totally. Yeah. So, well, Robert, thank you so much. I'll let you move on with your day. Yeah, same here. Thanks. I appreciate it very much. And thank you everybody for, for watching this and joining me. This, this show can be found in an audio podcast too, so if you want to let your friend know that this is available, I think this was a pretty in-depth conversation about the impacts of robots and AI in people's lives. I'm not sure that there's a lot of other shows that have dived to the level that delving on this. And, you know, and Robert's a terrific resource for this.

Yeah. And so, so I'm gonna— it'll be interesting to see how many people give, give this show to their AI. You know, it's like 2 and a half hours. Hey, but watch the 2 and a half hour show and tell me if there's any, anything interesting in it. Yeah. Yeah.

Or give me a synopsis, you know, just tell me the highlights. Totally agree. Yeah. So yeah, a new world. Yeah. And, and definitely leave some comments and questions. I'd be happy to answer them, and I'm sure sure Robert would love to see those as well. Yeah. And you know, you're always looking for, for feedback, and you're always doing, uh, Spaces sessions too, talking about these topics as well.

Well, I'm doing different things, and I joined a Tesla group, I joined a lot of AI groups, and then there's— on Thursday there's an AI show for developers, engineers, AI engineers, right? Nerdier. On Monday there's another guy who does a show AI news, right?

So yeah, popped in and out of different shows. Yeah, that's, that's fantastic. So I want to say goodbye to everybody out there and thank you so much for spending so much time with us. If you managed to make it through 2.5 hours, and this whole show is going to be on replay on LinkedIn and X as well as Facebook and YouTube. So, so yeah, then it's also available as a, as a YouTube or as a podcast, audio podcast. Trust Factor is the name of the show. So go subscribe to that.

So thank you so much.