0:00I had to decide, do I want to spend the next 5 years of my life being a professional victim, whining about how ruined I am and how I'll never amount to anything because of how poorly they treated me. You basically have to give them power over you. You have to admit, oh, I I have no control over my own life. The only people who have any control over it are Facebook, the people who bought my company and made it work.
0:21And so I I had to make a choice. Was I going to be a professional victim? Or was I going to start a new company and prove that I'm not a one-hit wonder?
0:28That I'm not the dummy that people said that I [music] was, and with the benefit of perfect hindsight, I made I made I made the right choice.
0:42When Palmer Lucky was a teenager, he spent his nights working in a camper trailer parked outside his parents' house in Long Beach, California. He built prototype after prototype, sometimes holding them together with duct tape, eventually he created something that became the prototype for the Oculus Rift. In 2014, when Palmer was 21, Facebook bought Oculus for
1:12He had built a machine that allowed people to leave the real world and enter another one. But just a few years later, Palmer was out at Facebook and he began building something very different. His company Anderil is challenging the giant legacy defense contractors and changing how future wars will be fought. But his work raises real questions. Can
1:42autonomous weapons make war more precise and protect lives? Or will it just make killing easier and more distant? If an artificial intelligence system identifies a target, who's responsible for the decision to strike? And what happens when wars begin moving faster than human beings can think? I'm Walter Isacson, and this is the builders, my series of conversations
2:11with the most consequential minds shaping our future. Here's my conversation with Palmer Lucky. Welcome to the builders. Thank you for having me.
2:21Hey, you were homeschooled by your mom.
2:24Tell me how that affected your emotional and friendship education.
2:29Oh, that's an interesting one. Um, I mean, friendship wise, I had a lot of friends who were homeschooled. I had a lot of friends who, uh, you know, were in the neighborhood. I also did a lot of when I was younger, um, you know, sports. And so, I I'd say, uh, friendship wise, it worked out perfectly fine. Education wise, I also think it worked out pretty well. When you're 14, 15 years old, you're taking all your high school credit at the same time as you're taking your college credits. Uh all of a sudden, you have the power to manage your own time. And you realize if
2:58I can be efficient in learning what I need to, efficient in passing my tests, efficient my homework, then I've got all the rest of my time to work on what I'm passionate about. In that my case, it was uh virtual reality technology. So I it's very obvious to me that Oculus, my first company, wouldn't exist if I had not been homeschooled. To what extent were your friends virtual friends and how did that change things?
3:21When I was younger, I had lots of in-person friends and then it was as I got older uh probably the the the people that I was closest to were people that I knew through the internet and through my online uh community for electronics uh enthusiasts called Mod Retro, combining modern technology with vintage computers, vintage game consoles, things like that. Interesting fact, when I started Oculus, most of the early hardware engineers were the other moderators and administrators of the Mod
3:50Retro forums, which I started. And and I were the people that I knew the best, that I trusted the most, and I I knew they knew their stuff on the electronic side, too.
3:58And you got this guy Chris, right, who's you'd never really met before in person.
4:04Yeah. Chris Dykus was uh the first employee of Oculus. Um, I gave him a call on a Friday and I said, "Hey, Chris, what are you doing next month?"
4:14He said, "Oh, I'm uh well, I'm uh I'm getting a job at this local pizza place and I'm I'm studying to uh you studying for these courses I'm going to take. I'm going to start college in the fall." I said, "No, you're not. You're going to come work with me and we're going to build virtual reality headsets at Oculus." That was what I was doing when I was a teenager and most most kids would not get the opportunity to do that. Elon Musk was really affected uh by reading sci-fi and certain comic books, superhero comics. What did you
4:42read as a kid that uh affected you that way?
4:45I was a big fan of uh of of science fiction that really focused on the technology itself. I guess one good example is is Jules Vern. Ju Jules Vern wrote 20,000 Leagues Under the Sea. And it's very obvious reading it that his goal was not to tell a story. It was to create a setting through which he could talk about what he thought the future of maritime technology was. I mean, you're talking, you know, a dozen pages in a
5:13row detailing the battery technology of the Nautilus and and and how it was recharged by extracting mineral elements from the seafloor and how how how how Captain Nemo was processing them. It was kind of a delivery vehicle for, you know, maybe maybe three dozen different ideas that he had and and and he wo a narrative around them. Um, I've always loved that sort of story because it it it lets you think what would the world look like if we had things we don't now.
5:41It was the same reason I I liked Star Trek even though it's communist.
5:45Uh, but was there anything in sci-fi that you said that's cool, let me do it.
5:49It kind of goes without saying it was the concept of virtual reality. Uh that was one of the technologies where it was uh it really grabbed me. Uh it was it was it was so interesting because it was so clearly not realized. But this idea that you could build technology that would trick your own peripheral nervous system into believing that it was in a place that it was not. And basically that you could you could build a mirror image of a person where a person is taking in data. If you could build this perfect stim, you know, simulator of
6:18that data so good that it could convince your lizard brain that you've been transported into a digital world of your own or someone else's creation. Uh, I mean, that's that's a that's a really entrancing idea for a technologist. And so, it is a lot of fun to be around at that moment where technology advances to the point where you can turn something that was science fiction into into science fact. And that luckily I was born at the right time for that to be the case with virtual reality. I could have been as passionate as I wanted, but
6:48had I been born 20 years earlier, I never would have made Oculus work.
6:51I have a theory about advances in technology is that the greatest advances are those that are the human computer interface or the human machine, how it puts together and VR seems part of that process, maybe culminating with neural link and chips in our brain. Do you see that as part of the main goal of technology is how to interface with humans or do you see it as something that might run off without us and not need us?
7:20To be honest, I think the entire industry's gone in completely the wrong direction. I think that we have it's it's been a folly and uh Steve Jobs, love the guy, but it was the iPhone that started all of this. Prior to the iPhone, there was an understanding that technology was something that could multiply your output that could that you could leverage it to work in superhuman ways. And Steve Jobs believed this. He wanted you he called famously called a
7:48computer a bicycle for the mind. Um you to allow you to accomplish things much more efficiently, to cover much more ground than you would unaded. The problem with the iPhone is that it simplified everything to a degree where yes, you could learn to use it in less than a minute. You could use learn to use it to the fullest of its capability in about 10 minutes, but then you reach this a ceiling. The problem is that everything has become the iPhone. Every
8:16application is now the iPhone. Every product is the iPhone. Every virtual reality headset is the iPhone. They are systems that are made to be easy to learn in the first minute.
8:26mastered within 10 minutes. And there's no there is no provision made for the man who will spend a thousand hours, 10,000 hours, 50,000 hours learning his art. There's nothing more to be had. And so personally what I would love to see is a return to applications and also input devices and human tocomputer interaction schemes that combine the two that say you know what this this scheme
8:53will require months maybe years of training to become truly proficient but what you get out of at the end will be something that is superhuman to an ultimate degree. I've got some ideas around that. But I'll even say things like Neuralink things, certain brain to computer interfaces. I see the same mistakes being made. Unfortunately, like I said, Steve Jobs ruined everything with the iPhone. We're still suffering the impacts of his incredible product design and uh I I'm hoping it comes back around someday. Well, let me push back on you a little bit because when you
9:23created Oculus Rift, it was indeed in the tradition of a Steve Jobs, it's totally intuitive and you just interface with the computer.
9:33Absolutely. I think it doesn't contradict what I'm saying. I just realized I was part of the problem. I made a product with a very low barrier to entry, very low skill ceiling. Anyone could use it. Uh, and then I sold a lot of them. And I think it's because of that, what's the right way to put this?
9:50I can recognize that it's easier to make a successful product when you follow the iPhone model. And then I can still lament the impact on humanity of of that because what a world we could live in if people were willing to invest time into their computer interaction interactivity skills the same way they invest in uh let's say sport or dance or or or piano.
10:14I think I think you'd see some incredible things come of it even if it wasn't for everybody. When it came time to sell Oculus, I know you want that, eager to do so, but you had a few choices. You had Microsoft, you had Google, you had Facebook. Why was Facebook the more obvious fit?
10:30The thing you have to remember about 2014 when this decision was going around is that all of these companies were in very different places than they are today strategically.
10:43And at the time, companies like Google and Microsoft had 10-year plans that really didn't involve VR. In other words, they were not betting that VR would succeed or certainly not betting that it needed to succeed, right? It was it was another part of a gigantic empire. But, you know, I I like Microsoft. I'm a huge Windows power user, but I recognize that the thing Microsoft needs to succeed is Microsoft I have Office, Microsoft
11:10Windows. It's it's not virtual reality as this new computing medium. Uh and the same was true of Google. Now Facebook was in a different position. Facebook didn't have their own operating system.
11:20They didn't control their own hardware platform. They were completely subservient to the people who did control these technology platforms. And so the pitch from Facebook was, hey, if you buy, if you sell to these other companies, they're just going to, you know, use you as another business line. And if you make money, great. If not, they'll take you apart for pieces. Whereas Facebook needs you to succeed.
11:45If we buy you, we are going to invest as much money as it takes to jump from this current computing paradigm into a future where augmented reality and virtual reality are are the future. And there's also kind of this belief you if you've got Facebook on the bottom of the ecosystem and you kind of shake up the whole ecosystem with a new computing revolution, maybe Facebook doesn't come out in first place, but they're probably not on the bottom, right? They probably move up a number of places in the process relative to Apple and and and
12:14Google and Microsoft and the like. And so their their pitch was specifically if you sell to us, we will put $1 billion per year into research and development for the next 10 years guaranteed. And when you start running the numbers on that and how much of your company you would need to sell to raise 10 billion to money to R&D, you you realize that's a that's a pretty incredible deal as a technologist. And and the funny thing is over the next 10 years, they didn't
12:42spend $10 billion. They spent about $70 billion. And so it the the theory has has proven out. They really do believe in this. They do want it to pan out.
12:52They need it to pan out. And I think that if we had sold to Google or Microsoft, that isn't the case. I mean, I don't want to throw Google under the bus here, but Google shut down their VR program about a year and a half after we were acquired, right? They were competing in VR. They decided not to.
13:06Then they restarted it again a few years later, and then they killed that one, too. And so it it's I I I think that if Oculus had been purchased by Google, nine times out of 10 we we we would have been shut down and knock on anyway.
13:18Yeah. But Meta just recently laid off 10% of its um virtual augmented reality force. It and they don't seem to be making it into a great consumer product.
13:28So there's two things to unpack there.
13:30one is great consumer product and I think first of all most of the money they're spending is on like research and development of nextgen AR VR technologies that that you don't necessarily get to see the the fruit of that until it eventually makes it into products. They've spent many billions on things that are important and will be important. Things like fundamental understanding of how to build certain types of optics, how to build certain types of cameras, build certain types of computing systems, and and you just don't get to see that until it gets into
13:59a product. Now, like you look at things like Meta Horizons, that's a terrible product. Everyone agrees awful product.
14:05I think even the people who work on it know it's an awful product. And I'm I'm I'm teamed up with Meta and I don't think that they even mind me saying it's it's a pretty bad product. Um, but I don't think you can look at that product and the mistakes that were made there and see it as represented the whole thing. You're right. They did just lay off about 10% of Reality Labs. But I can't see that as an indication that they're abandoning VR because well, two things. One, they've still got 90% of it. And that 90% is literally 10 times bigger than the next biggest VR team. I
14:35mean, so I they are larger by an order of magnitude post layoffs than the next closest comp competition. And so when I see people saying, "Oh, they've lost faith in VR. They're moving on." My point is, how how could you say that?
14:47They clearly still believe in it more than anybody. They've spent more than anybody. They've made an adjustment where they laid off people. And remember that these jobs are focused in a few specific areas. For example, they've decided that they are not going to be in the business anymore of building firstparty gaming applications that compete against other developers. So, they said, "Wait, okay, we're not going to be in the business of competing with Ubisoft and Rockstar and uh Capcom.
15:12We're just going to work with them instead of kind of being the platform we're trying to attract them to and also competing with them on first party titles." And so, yeah, all the people who are doing that got laid off. And I feel bad for people in that position because, you know, it's their livelihood at stake. But at the same time, recognizing that I feel for those people, I can think it was the right decision for the virtual reality ecosystem. I think it'll make it stronger in the long run to not have the biggest VR company competing with the
15:40video game developers that they need to attract to their platform. Because there's a lot of developers who say, "I'm not even going to develop for uh the MetaQuest. Why would I? I would be I'm competing with their own games. They're going to put their games in the top places on the store. They're going to make all of the ads for their games.
15:55They're going to have cheaper, more subsidized pricing. How am I going to make money competing with that?" This, I think, ends that problem mostly.
16:02Wow. You seem pretty passionate about it. Let me ask you to reflect. Do you kind of wish you were still at Meta?
16:10I think kind of wish is is the right way to put it. Uh I really enjoy what I'm doing at Andreal. I think that it's making a huge difference. I think it's making a bigger difference than I would currently be making in the VR industry.
16:22You also have to consider like it's more than just do you wish you were still at Oculus? Because if you were to put me into Oculus, it wouldn't solve a lot of the underlying problems. for example, most people in the Bay Area uh are very very much on the other side of me politically and they hate me and they don't want to work with me. And so there's an interesting you it's like do I like do I really do I wish I were back? No. I I really wish the whole world were different, everything worked differently that we could rewind time
16:51and that I were back at Oculus.
16:53Well, wait. First of all, you say the whole tech community is against you politically. It seems there's been quite a shift in your part of it in Peter Teal and David Sachs.
17:03So that that is fair. There has been a bit of a bit of a shift and so my whining is a little bit out of date. Um but but but certainly for a long period of time you you kind of have to ask yourself do you want to do you want to do you want to swim in a stream against the current or with it? I mean some I like being contrarian but I like being in a stream where I can get as much done as possible.
17:24So let's go back to the firing. uh you got fired basically because you contributed nine or $10,000 to a Trumpup supporting pack. Then they said, "Oh, no, that wasn't the reason. It wasn't political." But that was totally untrue. Do you wish you had pushed back and fought for your rights? The real answer is yes. Um you got to remember I was 24.
17:47I was a young guy. I didn't have experience in any of this. I I think I probably I probably should have fought harder, but you also have to remember that once you've been fired, you can't you can't fight your way back into being in charge of your company that you started as as a as a teenager. You can't sue them into making you the CEO, right? That that doesn't happen.
18:12What you're actually doing is just litigating the past, maybe getting some money for the future, but then in the process, you have to put so much time and energy into that that singular effort. It's actually worse than that.
18:26You can only get the money if you can prove damages, right? You have to prove they damage your reputation. Typically, that would mean you'd want to prove that you they ruined your ability to raise money to get people to work for you. You basically need to convert your life to win these lawsuits into that of a professional full-time victim. And so I had to decide, do I want to spend the next 5 years of my life being a professional victim, whining about how ruined I am and how I'll never amount to anything because of how poorly they
18:54treated me? You basically have to give them power over you. You have to admit, oh, I I have no control over my own life. The only people who have any control over it are Facebook, the people who bought my company and made it work.
19:06And so I had to make a choice. Was I going to be a professional victim? Or was I going to start a new company and prove that I'm not a one-hit wonder, that I'm not the dummy that people said that I was? And I think in in in with the benefit of perfect hindsight, I made I made I made the right choice. In a perfect world, I would love to have fought for what's right and also started Anderl, but it's hard to imagine that I could have done both effectively simultaneously.
19:32Well, wait. Could you imagine yourself being like an Elon Musk and both running the uh augmented reality at uh Meta and also Andre or do you think you have to focus on only one thing?
19:44I think Elon is a singular individual.
19:47I've never seen anyone like him at all. He's he's he's the only person I've seen effectively do and people like like what's the next best step? It's like Jack Dorsey with Twitter and Square.
19:59Steve Jobs with Pixar and Apple. That's pretty good. But, you know, he's dead.
20:04So, it's it's a it's a it's a but fair enough. But, look at Elon. He's doing it with SpaceX and with Tesla and with uh now X and X AI. He's a truly singular individual. I don't I don't know. I don't know anyone who can do it the way that he does it. I certainly don't think that I can. And I've given it I've given it some thought. Um, you know, I have a couple of these little side projects like Mod Retro, the online community, is now manufacturing clones of vintage game
20:33consoles and and like we we build a tribute to the Nintendo Game Boy. It's basically a a Super Game Boy with a sapphire screen lens and a and a and a magnesium aluminum alloy shell. It's the fanciest Game Boy you could ever make. The best experience you could ever make with a perfect clone of the Nintendo Game Boy Color LCD.
20:51But like that that's a tiny tiny slice of my time. I also am a major part owner of a niche ice cream company that makes ice cream burritos. I put very very little time into it. And is taking up 99% of my time.
21:05Well, let's get to Andre then. You're sitting there. You're fired from uh Facebook. You got a lot of options. Tell me the path that gets you to be a defense products company. Well, this is something that I've been thinking about for a while and a few of my friends have been thinking about for a while.
21:23We've been talking about the fact that the United States had been falling behind in defense technology. We were no longer building the tools that we really needed to deter or win the wars that were likely to happen in the future. and the incentive structure that the United States government had created for the major defense companies clearly was not incentivizing them to invest in this type of technology. Simultaneously, the big tech companies were refusing to work with the national security apparatus.
21:51And there had never been a point in US history where that was the case. There had never been a point where the most innovative technology companies refused to work with the United States military.
22:02That's always been a hallmark of American uh technology companies is that they did work handinhand with the government on this to safeguard the freedoms and structures and uh even trade environment that made it possible for them to operate here and internationally. And so I said, okay, if the big tech guys aren't going to save us and then the defense companies aren't going to save us, well, maybe there are some defense startups out there that could do something. And so me and I I
22:31didn't start out wanting to start a defense company. I started out wanting to give money to other defense companies. I wanted to invest in them and help them out using the money that I'd made. I actually started doing this while I was at Oculus. That was how I got connected with Trey Stevens at Founders Fund because he was at Founders Fund tasked with finding the next SpaceX or Palunteer, the next company worth billions of dollars doing business with the government. He had met with I think about 2,000 different companies. And how many investments do you think that they
23:01had decided to make? Zero. It wasn't that these were bad people. It wasn't that they weren't patriotic or even smart. It's that it was very clear none of them had the set or constellation of skills, ability, charisma, whatever it was to be able to build a company that could go toe-to-toe with Lockheed or North Northrup or Boeing and actually win. In other words, to go from being a little gizmo and gadget provider into being a next generation defense prime.
23:28And so when I got fired, Trey was one of the first people I called. I said, "Trey, I think I think we need to stop trying to invest in a defense company that's doing this. I think we need to start it ourselves." And he says, "But Palmer, what are you going to do? I I thought you're too busy with Facebook."
23:43I said, "Ah, well, it's a sacrifice I'm willing to make." [laughter] It worked out pretty well. But but you know, the whole point of the company from the beginning was to save taxpayers hundreds of billions of dollars a year by making tens of billions of dollars a year. So to build things that are much more efficient, much more cost-effective, and to do as quickly as we can and get it out there as as quickly as we can, and we've done it.
24:05Well, tell me what type of big systems that you can do better than Boeing and Lockheed. Well, you know, I I don't want to I don't want to be too uncou and and you know, throw people throw people under the bus, but I mean, one obvious example, not judged by me, but uh judged by the US Air Force was uh our recent success on the Air Force's CCA program, the collaborative combat aircraft program, basically building AI powered fighter jets for the United States Air Force. our aircraft Fury which now has
24:34the Air Force designation FQ44. We beat Northrup and Boeing and Lockheed for the CCA increment one contract. I I think that's going to be the first of many. I think there's been other smaller things where we've won here and there. You're going to see us winning major programs, I think, on a regular cadence because investments that we've made over the last eight years are finally bearing out and people are going to start to to see the fruit from those trees. It seems
25:03that you know, Alex Karp with Palunteer, Elon Musk with SpaceX and you have this different model where you're not getting commissioned by the government to do a certain thing. You're creating a product and then saying this would work for you.
25:20That's right. So, we use our we use our money that we raise and that we make to decide what products we're going to build, build them, decide when they're done, and then sell them to the customer. There's times where we're developing handinhand with the customer.
25:33And there's certainly times where the customers are helping are helping with cost because of some particular thing they want to do, but generally speaking, Andre is a product company, not a contractor. And we put billions and billions of dollars into things before the government ever gives us a dime. And we're not we're not talking about a few months here. I mean, we have aircraft we've been developing for five years now that we're just now for the first time getting any money from the government for. And uh it's it's I I think it's actually a much it's a much healthier
26:02model. It keeps you a lot sharper. It ensures that you skin your knees when you trip and fall. And that's what keeps you on on your toes. Let's get to the name Anderole which of course from Lord of the Rings but it's the force for good sword. Do you consider yourself having sort of a moral component here that you're going to be fighting for say democracy and American interests?
26:25Yes. I think that the one of the ideas with Android is we are not a politically neutral company uh a morally neutral company an ethically neutral company. We want to be building a company that is working with the nations that safeguard rights like freedom of expression uh freedom of self-governance uh these kind of ideas that you know maybe not every country perfectly agrees on. We're not always going to you we're not always going to be perfectly in sync with each other but we generally more or less agree that those things are important and there's a lot of nations in the
26:55world that don't think those are important. Um at the same time I do recognize at at the end of the day we're kind of beholden to the United States government. I can't sell to any country the United States government doesn't want me to sell to. Um, and any ally that they want to work with, we also have to work with as well. I I've I've decided that that kind of is the way that it has to be. Uh, regardless of my moral position or ethical position, I live in a country where we've decided that we are going to outsource responsibility for those decisions to
27:24our constitutional republic, to the elected officials that we can all hold accountable for those decisions. And so I sometimes get people saying,"Well, Palmer, does your commitment to ethics extend to, for example, uh saying that you will help one country or won't help another?" The point I have to make to people is you don't want to live in a world where a handful of corporate weapons executives get to have de facto control over US foreign policy. That that should not be a decision that executives like me are able to make.
27:53That the only way to keep us accountable is to make us an extension of the United States government's policy. Otherwise, you end up with very very strange power dynamics where you can imagine a world where the CEO of Loheed Martin has more power than the president of the United States and that's probably not a good thing.
28:08Tell me about the software uh platform Lattis and how that sort of underlies a lot of what you're doing.
28:16So, Lattis is the collection of software that we've built that powers all of our different products. It's kind of the AI brain that powers our missiles, our submarines, our fighter jets, our sensor towers, all of the 25 plus products that we make. And the really key thing about it is that we can tie together all of these systems so they can talk to each other, work together with each other very seamlessly so that every sensor is an eye for every other system and every
28:43weapon is a system available to every brain. It's a very different way of thinking about autonomous systems to to kind of tie them together in this way.
28:52Uh and it also is great in terms of a return on investment perspective for me because if I put a dollar into making the wing of an aircraft better, the only thing that gets better is the wing of that one aircraft. If I put a dollar into making lattice better, everything that uses lattice gets made better. And so you this this mult multiplicative effect where putting billions of dollars into it now turns into billions of dollars you've invested across the
29:19dozens of products that use it. And uh we actually have as many people working on Lattis as we do every hardware product combined at Andreal. That's how important it is.
29:29And is the point to make autonomous weapons?
29:33Yes. And do you see there's any problem or maybe there isn't between uh by having drones or autonomous systems that will make kill decisions without a human being in the loop?
29:46I think it is the only rational moral outcome or decision to make. Life and death decisions are too morally fraught and too important to not bring to bear the most powerful technology that you have. And this isn't even an argument about AI or not. It's a AI happens to be the thing that everyone's talking about right now because it's hot and it's in the news, but what this really comes down to is do you want to use the best technology that
30:13can measurably and provably lead to the lowest collateral damage, the lowest loss of civilian life, the highest level of precision. There are people who would be willing to accept more collateral damage, more civilian loss of life in exchange for autonomous weapons not being allowed. If you had an autonomous landmine that could differentiate between a Russian tank and a school bus full of kids, there are people in the world who seriously argue that it is
30:42better to have a dumb mind that blows up that school bus full of kids as long as it wasn't a robot who decided who lived and who died. I just can't abide by that. I I I cannot I cannot get on that side. And people will point to problems like, well, what if it makes this type of mistake? What if it does this? What if it does that? And my point is, let's regulate that. Let's regulate the the quality of these algorithms. Let's regulate their output. Let's regulate how they're used. Let's make sure that we're holding people accountable for when they're used or misused. But when people in the United Nations, for
31:11example, call for a ban on what they call lethal autonomous weapon system laws and they say things that the sound bite sounds good. Well, don't you agree that a robot should never decide who lives and dies? It sounds great as a sound bite, but what it actually turns into is that you're committing yourself to making dumb weapons that cannot discriminate between actual combatants and civilians. It's it's it's weapons that cannot differentiate between the right building and the wrong building.
31:40Well, wait, correct me if I'm wrong, but what you've just criticized is the US military doctrine. I mean at places like Cree Air Force Base, they got drones flying all over the Middle East, but there are humans sitting there watching it and they've got to be in the loop. Do you think the US military doctrine is wrong? I think that their doctrine is formed by the limits of the technology that they have. So I think that right now it is good doctrine. Up until now, the level of autonomy that we've had,
32:08the level for a machine to make that decision has been so incredibly coarse and poor quality that it would have been a really bad idea to just generally hand it over. That's why I said this isn't really about AI. It's about, you know, is it the right tool? You also got to remember autonomy in many ways is getting rid of failure modes. Um, so like you you've probably heard about some of the people hacking into predator feeds that were unsecured. Think about all the problems that go away when you have an autonomous system that isn't even carrying a radio because it doesn't need one or use one for the mission.
32:37Like you're you're getting rid of cyber vulnerability in a lot of remotely piloted systems. That same requirement to have somebody press the final kill decision button can also turn into the a way for every weapon to be hijackable by an enemy who will then tell it to do the wrong thing. And so I think that doctrine is formed by the weapons that they have. And and what you said is mostly true. Like it's true for a predator drone. A person has to be in the loop. They have to see what's going on. They have to push that button. But
33:06the United States has been using radiation seeking missiles since Vietnam, right? You send a missile over the horizon, you never see it again. It looks for a certain radio frequency emission and it flies into that thing.
33:17But the United States has a lot of weapon systems that already take kill decisions into their own hands. It's just they're very specific kill decisions. Uh and so I think AI will be the same way. Well, I mean, you're right. Landmines are completely autonomous.
33:32Well, now now you're getting into a whole another uh interesting debate. I I gave a talk a few years ago which argued that autonomous weapons are actually thousands of years old. So, landmines, great example. They are an autonomous weapon. There have been landmines for many years that ignore certain types of targets and attack other types, mostly based on like weight over a pressure plate or their magnetic signature. So, you have a system that is making decisions about who to kill and who to not. They say, "Hey, we're going to ignore cars, but then a big armored
34:00vehicle is going to have a big a more a larger uh fair, you know, ferrris uh ferrris mass. It's going to have a larger magnetic signature, and I'm going to blow that up." But if you really want to go back, you can actually find things like pit traps that were built the same way. Pit traps that would allow people to walk over them, but horses to suddenly break through and fall through so that they wouldn't get the foot soldiers, but they would get the higher echelon officers that were on horses behind them. So people have been building weapons where they were designed by a man using the power of
34:30technology to decide on the battlefield who will live and who will die at the point of that weapon. Uh I I don't think there's actually an inherent difference between using AI to do it and using how thick the bamboo over is over your pit trap. They're both two sides of the same coin. The key is that a person has to be accountable for the use of force. A person has to be accountable when they're misused. and the people who are building them and using them have to be aware of the fact that these things have problems. You know, like if you build a pit trap in the middle of a in the
35:00middle of the wrong street, you know that you're going to be held accountable uh for for for what happens next. You don't get to say, "Oh, well, I didn't I didn't I didn't mean for the pit trap to kill the wrong person." I think AI weapons will be the same way. Can you either in your science fiction brain or in your andal brain imagine a war in the future that's fought almost virtually by really smart machines in which very few
35:26people if any die? I mean, I've had a lot of discussions on this and it's it's it's hard to it's hard to make the point I'm going to make with without people jumping down your throat and trying to to take it in the least charitable light. Uh but I would say we're basically already living in the early version of that. You you look at modern warfare uh and a lot of a lot of the wars that have happened over the last call it 30 years a lot fewer people are
35:54dying in war. Generally speaking we have like the invention of precisiong guided bombs has made a lot of old school bombing campaign techniques completely irre relevant and and seemingly barbaric. Right? The things that the United States and our allies did during World War II today would be considered war crimes because we've become so good at precision targeted warfare. The reason people jump down your throats, they say, "Palmer, how can you say, you know, a war is humane or that nobody dies when you're talking about tens of thousands?" Obviously, that's horrible.
36:24But remember that in living memory, we have wars where tens of millions died. I mean, like, look, look at Russian losses. Look at Chinese losses. Look at German losses. just an incredible loss of life. And so I think we've actually already moved about 90% of the way down down the scale. We're kind of fighting over the margins here. Like will wars go to where, you know, there's there's very few people involved and it's mostly machines. I would actually argue we're already most of the way there. This is why I don't think it's actually going to
36:53make much of a difference. People say, but Palmer Marino, the future of war, you wouldn't that mean that the barrier to war is going to go way down? And my point is, guys, we're it's it's already there. war wars just don't carry the same collateral damage that they used to. Um and so I I very much can imagine a world where you won't get rid of all de all the human death. I think we're going to get to a point where most of the casualties are going to be uh robots and and systems and hardware
37:23rather than flesh and bone. Uh and I I I think it's it's probably a good thing that we're already 90% of the way there.
37:30Well, don't you think your technology will end up deterring more wars than it creates?
37:36That is certainly the hope. I believe that it's an interesting thing where the more successful Anderol is, the less credit we'll get for it. And so there already today you have people saying, "Oh, but Palmer, oh you Palmer, Palmer thinks that we're going to fight this war with China. Well, sure doesn't seem like anything's happening yet. And the funny thing is, let's say that we get to 20 years from now and we haven't had a fight with China. That's going to be in large part because of not just
38:04Anderol, but dozens of other companies that have all contributed to this.
38:08Dozens of countries that have strongly contributed to that. The only reason that we will avoid it is because we carefully build a world where China doesn't feel that they can move. The annoying thing is that everyone will look at that and assume it's just the natural state of things rather than a carefully engineered beautiful garden and uh that's that that get that gets a little frustrating. On the other hand, if we do end up in a war with China and Anderil systems are getting used on the regular, all those people say, "Wow, I
38:38was wrong. Good thing we have Anderil." It's like at that point we've already lost, right? by by the time by the time you're by the time you're uh fighting a war, you've you you it means you didn't build your garden correctly.
38:51One of the other people we having on this show, the builders, is FA Lee, whom you know, and she said, uh AI is a powerful tool that we're only beginning to understand and that's a profound responsibility.
39:05What are the responsibilities right now we should figure out in this tool of AI?
39:11I mean, the number one the number one thing I'm most worried about and thus I think has probably the biggest area of responsibility. Like I'm I'm I'm I'm not a pro censorship guy. I'm not a fan of limiting AI to not do certain things.
39:24But if I were going to pick one area to have some level of responsibility or regulation or censorship, it would be around applications of artificial intelligence to create biological and chemological weapons. There's so much lowhanging fruit that has been inaccessible to people because it was buried in academic understanding that required a certain bar to even be barely functionally literate in. Uh AI has made
39:54it possible for people to get much further down the chain of doing very scary things. And you you've probably seen some of the things people have been doing where they have uh labs that are doing um uh uh new pharmmological development where the AI is basically telling them which experiments to run, which compounds to mix, exactly what their work instructions for the lab are.
40:15And you basically have like quite re you know you you don't need like PhDs like you just need like lab technicians who are reading their work instructions and doing these things, running these tests, feeding the results back into the AI. They're basically using the people as meat robots to do everything for them.
40:30We are not far away from a person being able to create pretty scary biological weapons without any of the people who are working in the lab knowing that that's what they're doing, without any of the machines catching it. And I in terms of responsibility, if I I I I think that there would be a very strong argument to remove certain elements of data or specifically tamper with the data in order to break it unless you are within
41:00institutions that that are allowed access to these things. And I people say, but Palmer, you know, these tools are also useful for developing medicine. You know, I I think that it's an asymmetrical thing, though. I agree.
41:11This would mean that probably you won't have a teenager in his garage able to develop a new far pharmaceutical compound uh without being part of a laboratory. And I I that that is that is a loss that I'm willing to take if you don't have people designing bioweapons designed to wipe out people based on their ethnic line, race, uh other other biological markers, or thinking they're doing such and and failing to do so. I'm terrified of that.
41:38So, if there's going to be responsibility for AI, I think we should focus on things that can that can uh self-replicate and live in the real world outside of that data center.
41:47All right, one last question. I'm going to ask you to pick something that you think will be really cool in the future that uh you'd like to be doing. What What's What's the next cool thing?
41:59Oh, this is so difficult. I only get one. I just get one pick.
42:02Oh, two. All right. I think one of the things that I'm very excited for is the vertical integration of power generation, manufacturing capacity, processing capacity, and recycling capacity where you are going to make it, I think, ridiculously cheap to build a lot of goods. The same mechanization you've seen in textiles manufacturing where fabric is now roughly a thousand times cheaper in
42:30terms of per unit earning cost as it used to be. I think you're going to see that applied to cars. You're going to see that applied to consumer electronics. I think you're going to see a lot of everyday goods because the extraction, processing, and recycling cost, manufacturing cost, it'll all together be so low that we're just going to be able to crank out things like TVs and cars as easily as we churn out rolls of fabric. I I think a world where you can buy a car for $1,000 is something
42:58that we're, you know, adjust for inflation however you want, but I think that's actually within our lifetime. So, I'm very very excited about that. Just the basic necessities of life being driven way down in cost through automation, vertical integration, and collocation. The second would probably be settling the oceans. There's there's a lot there's a lot of ocean out there.
43:17There's a lot of space. There's a lot of food and energy potential and space potential and up till now the main reason we have not been settling the oceans is it's it's been a it's been an engineering limitation. I think that new advances in material science make that not so much of a problem anymore. And so I think we're going to get to the point where we're able to build mega structures at sea that persist for thousands of years without any significant maintenance. And that'll
43:45that'll that that'll be a very very interesting thing that'll uh that'll that'll that'll reorder the economy and politics in a lot of ways.
43:52This coming from the kid who read Jules Vern, right? [laughter] Look, I I got to say uh the best way to build the to predict the future is to build it. But uh I think people often end up building the future that was put right in front of them by science fiction authors. And so I I I can't say I've ever come up with an idea that some science fiction author I've read hasn't come up with better. But uh yeah, I'm I'm a big believer in the concept of seasteading. I think it's going to come back around. It's it's out of vogue
44:20right now, but it'll come back. Palmer Lucky, thank you for joining us.