Palmer Luckey: Important Development in AI Weapons

Why tech doesn't work with the military

00:00For the first time in history, the most innovative technology companies in America are largely not working with the United States military. That's truly unprecedented. There's never been a point in US history where that was the case. The most innovative technology has always been accessible to our military and the most valuable companies have always worked with our military. And I think that really gets to the core of why we're doing what we're doing. Uh Russia and China don't have these issues. Their best technology companies are working on defense challenges. Their best people are working on defense challenges. And so if we end up crippling ourselves by putting our best people to work on search engine

00:38optimization and ad delivery, how can we possibly expect to compete? And that that's what we're trying to build here is a defense products company, not a defense contractor, but a defense product company. >> Is that a sign of a decaying society that the best people won't even defend their own society or like what is it a sign of? Oh man, I I I like to think that it's actually just a it's like a specific trend of the of the modern time and strategic flows that are going after like this didn't happen by accident. I think China's done a very good job of making China does a great job of keeping the entire world corporate world on a hair trigger where they know that if they do anything that is going to upset

01:18China they won't get access to Chinese markets and they won't get access to Chinese capital. And we see this especially in the media with the film studios. You know that the idea of making an anti-China film in the modern day is unthinkable to most of these studios. And so uh I I think that it's not like a it's not just a sign of moral decay. In fact, I'd say probably eight out of 10 people working in the tech industry, if you ask them, do you want the United States to have the strongest military in the world? Eight out of 10 would give you a strong yes. the last you you know one of them that says they're not sure and one of them that would be against it. So most people in the tech industry I think do agree the

01:56problem is the corporate dynamics make it more profitable to keep China happy and to not piss off the media which is uh kind of disproportionately anti-defense. >> Fascinating. So in a sense you're like 2020's version of a hippie who's overcoming the corporate interest to push what push what the common people want. I like I hadn't thought about it way that way, but it is it is actually really really similar and like in the same way that I think you had like people people complained that we had a disproportionately uh you know pro pro- uh corporate and pro- mililitary media and I think that really has flipped. I don't think anyone would argue that let's say the American tech press uh or even just the the press in general is

Anduril's success and innovation theater

02:38too promilitary. I I think that idea you know went went away long long 100%. So let's I want to know more about Andre. So Andre Palmer like to me that you've hired like all this amazing talent. You have all these amazing projects you're doing. You're growing revenue faster than any defense company has in history as far as I know. A lot of people don't have this perspective. A lot of people they hear these big defense companies who are very threatened by you downplaying you talking about oh there's one or two products. It's not a real defense company. So there's a lot of people who are kind of who who want to tell a more negative story. I think it's because they're jealous and threatened. But give give us a sense of your success

03:12so far. give us a sense of what you're doing so people can hear it directly from you. >> Well, yeah. I think there are people who downplay us and then they also overplay their competition. You they say, "Oh, Adel is doing great AI stuff, but we have great AI, too. It's really, really good." And it's not. Uh, you know, but they It's really cheap to say words. You can just say, "We're as good as Anderol." And until you actually get to a real field trial, the words are as good as the actual >> that was the same experience with pounds here with me is that we are competing against a lot of innovation theater where they couldn't actually do it when they actually had to deploy it. It took years to deploy. It took billions of

03:46dollars and didn't actually work. >> Oh, I mean, you know, for people who don't know what innovation theater is, you know, there's there's all these people at DoD who have this job. Basically, their job is to generate things that look real enough to make a press release or to do a panel or to release a video or to talk to Congress. And none of it is actually real or useful or deployable. It's it's pretty wild. >> And so so so tell us about a few of your products. What are you guys actually doing? What that you could tell me about? >> Sure. So we're doing a few things. I mean the first product that we worked on is called a sentry tower. It is a solarp powered infrastructure independent AI surveillance tower uh that you can put

Sentry Tower and border surveillance

04:24basically anywhere. whether around a military base, a border, critical infrastructure like power plants, desalination sites, and it uses radar, thermal vision, visible spectrum imaging, and a few other sensors fed into a probability model running on a computer that allows you to detect all the people, all the vehicles, all the animals, all the drones that are a couple miles around it. And these things also mesh together. They talk to each other. So you can deploy, you know, strings of these things a couple miles apart and get really good awareness of everything that's going on around your base or around your airport or around your border, what whatever it is you're trying to understand. >> I remember when you were first building

05:02this and you had the thing that said like 97% horse because I don't think you entered a cow model yet, but that was that system, right? >> Yeah, we had some problems early on where Yeah, you're totally right. We had we only had one classifier for that type of animal. So, uh, you we're always getting the animals on. We're actually a lot better now, but usually we just report it as an animal. It's >> you don't bother telling them that's a what type of horse or donkey it is. You just say animal. >> Yeah. >> Yeah. Most of the time like it is the customers can set it up to be more specific in that, but generally they want fewer variables to sort. They don't want to have in their list of variables like uh, you know, all the types of

05:38horses and cows and donkeys and goats and then people. They just want to have animal like ignore animals. And you know that's that's generally the the right thing. It still keeps track of what went through, but they don't want to pay to attention to it in the moment. >> Um, >> and this is being used on a lot of the border. How much of the border is using this right now? How are they using it? >> I can't tell you, but I can tell you that there's hundreds of towers out there. Um, so, you know, just spread across a lot of different places. So, we've got a lot of these things out there. And one of the things we're really proud of is that we have over 97% uptime on these towers. uh which I think we can get better but most of that is

06:14down for like maintenance and upgrades. Compare that with our competition which is not AI powered requires people to be watching all the sensors. Like that's our cool thing. We have >> I heard that people were driving these pickup trucks where they would have a metal grate and then they'd look for footprints be the metal in the in the smoothed out dirt. I mean it was that bad, right? >> Yep. It's called cutting signs. They just drag stuff on the dirt to smooth it out and then they go back later look for footprints look at the direction and then they go and try to find them. I mean the the problem is that that's very labor intensive. You spend all of your time trying to understand what happened

06:45a while ago. And with our system, one person can understand everything that's happening across hundreds of square miles. That that's that's a total game changer when you're talking especially about military applications where people imagine you can just have an unlimited number of people, but it's not true. You can only have such a large footprint. You know, there's only so many people you can support at one of these forward operating bases. And so if you can take people off of useless tasks, put them to use on useful tasks, that's a that's a really big deal. Um, and then as far as other products, I think we we built a few things that kind of enhance the capabilities of our sentry to powers. Uh, one of them is Anvil, which is a

Dust sensors and coverage gaps

07:20counter drone interceptor that can knock other drones out of the sky. We built a lot of jamming technology and electronic warfare technology that can uh that can detect and destroy and jam a lot of different electronic systems, including drones. Uh we've built a small sensor. We can call it dust. It's basically a very small disposable sensor that you can carry a bunch of in a backpack. And it has a much shorter range than a tower, but it has a radar. It has a camera. It has a processor. It has a radio. And so you can basically just toss one of these things on the side of a tree, on the side of a hill. They you we have versions of them that look like like little rocks. And they're able to detect when people or vehicles are going

07:59by and immediately classify that and feed it. >> How how does a little rock work? Is it listening or is it watching or or what is it doing? >> So So it has a it has an onboard radar on that is putting out it's it's very low resolution radar. Basically think of as like a motion detection radar. It just detects when there things are moving when things are moving. It powers up the rest of the system and then takes pictures of it with thermal imagery and EO imagery. >> Oh, got it. So if a horse goes by, you'll get the picture of a horse, but if a cat if a car goes by, you'll get the picture of the car as well. >> Yep. And if it's a picture of a horse, it will usually just ignore it. Says, "Oh, that's fine. Turn back off." If

08:32it's a car, it remains on, continues watching it, and then starts a radio link back home. And uh it that that's a really useful tool where I don't think we'd be able to make money off of them if we were selling them on their own. But they make the rest of our product so much more valuable because like with sentry towers, you you have geometry problems. Like I can't see behind a hill. I can't see behind a building. But if you use dust plus sentry towers, you can take an install that was 95% coverage and turn it into 100% coverage, which is >> you might have towers and then you add some dust in different areas around and then you have all of a sudden you have everything going on. >> And then I think on the on the on the

Ghost drone capabilities

09:07kind of airframe side, we've been doing a lot more aerial vehicles. Um our our main one that people know about is called Ghost. It is a man portable electric helicopter drone. You carry in a backpack, folds up into a carry case that's smaller than a uh than a uh than a typical DJI. >> Give me some give us some stats on Ghost. Like what how fast does this thing go? How long does it go for? >> Uh so it'll cruise at about 85 miles an hour. It can fly for up to two hours on a battery and up to six and a half hours on a hydrogen fuel cell. Uh which clips on in place of the battery. Uh the really cool thing about it is like the airframe is incredible. It can carry a ton of payload. It can fly for a very

09:44long time. It's also a we call it ghost because it's basically inaudible and almost invisible. At 100 meters, it is inaudible to the human ear. Wow. And at it because because it's a helicopter that's very very skinny and very very short and we stack everything all in a line. When you look at it dead on, it's like it's like this big. Like this is the front of the thing. So try looking at a square that's like, you know, three inches tall and three inches wide. 100 meters away. It's basically invisible. And if you get a few hundred meters, like let's say 300 meters, you you are invisible. >> Does it fly if if it wants to be stealth, it flies 300 meters high then and just like goes over and people don't

10:20see it. >> Exact. Well, not even 300 meters. We can actually we can track people at about a kilometer. So, you know, a thousand meters we can be out and we're following people, classifying people, and we don't have to and they just never hear the thing. >> They have no idea it's even watching them. Could you put weapons on this thing or that's not a thing you put weapons on? So, we do not put weapons on it yet. Uh, but you know, I I think that I I I mean, this this is another whole discussion, but I don't think there's a moral high ground in saying you refuse to put weapons on things. I've seen a few other drone companies doing this, like, oh, we would help with this, but we would never we would never help

Weapons, precision strikes, and ethics

10:53deploy a weapon system. It's like, look, the the the like the US Army is a weapon. The US Marine Corps is a weapon. There's no high ground in in say, "Oh, I'm just the guy that found the target, pointed a laser at the head, and then let something else go after the guy." Like, if if if you can make something that allows for more precision strike, less collateral damage, I think it'd be morally reprehensible to refuse to do it to try and keep your own hands. >> What's an example of something that would be a higher precision strike? What's an example of of a of a new technology for higher precision strike? You can give us something you think about >> like the easiest example is just things with smaller booms. I mean like you know

11:30a predator occurring a hellfire that's a that's a pretty pretty big piece of ordinance and you we've seen a lot of unfortunate situations where a strike has caused collateral damage uh because it was too close and you couldn't make something that just takes out one person. They they've actually been do actually one of the interesting things on the hellfire there's a new version of the hellfire that uh basically just runs into people. It doesn't even have explosives. It just runs into targets and is has very little collateral damage. Um, I think an example in the future would be like what if you could make that into like what what if you could say I'm going to have like right now a very low collateral damage strike

12:05would be like let's say a US Marine uh uh you know sniper pair and that guy can take out an individual person without anyone around them getting hurt. If you could take that capability and push that forward so that it could be done from a remotely operated platform that doesn't require putting a sniper team up on a hill where >> the obvious thing would be is if you had something very high above that could drop something or something like that. >> Exactly. I mean there's a there's a million ways to to imagine approaching the problem. But I think that that's what the future of warfare is going to look like. Does it does it does it scare you to imagine the bad guys were competing against getting tools like

Adversary drone threats and defense

12:41these and being able to drop things very precisely on us? Let's just say let's just say they had this and it was hard to trace and they cited the Palmer that you were a high value target and they wanted to take you out. It seems pretty scary to me. How do we like how do we prevent that from happening? >> I the thing is they're already doing this. So, >> but they haven't t they're not taking anyone any of us out yet so far. Like they're not taking highv value targets out in the US hopefully. Well, I mean, like there are armed drones attack, armed drone attacks that are going on. They're they're doing real damage. Um, I think that that not all of this is necessarily broadcast to the public in a really big way. Like, if you

13:15get attacked by drones and it's successful. You don't even want the guy who launched the drone to know that it was successful. You don't want to basically announce, hey, this is the tactic that works. They got us. We had no way to stop. >> You want to make it look like a heart attack, but that's hard to do from a drone. Unless you guess you shoot through shoot tiny thing and it gives them a heart attack, huh? I don't know. you who who knows. I mean, that that type of stuff is going to happen. But even if you can't give someone a heart attack with a little pin, you can definitely shoot them or, you know, blow up a little tiny high explosive charge. And we do have adversaries that are doing that right now. I think they're

13:44going to do it more and more. And the key is you have to build the defensive system. Like that's why we're building anvil. That's why we're building our counter UAS systems in general. Um, >> but but they're doing that right now in theaters of war like Syria and Iraq. Are they doing it other places too or you can't say? So, I can't say any specifics, but we have done combat validations where we've sent these things down range. Yeah. And I mean, that's that's one of the biggest threats that our troops are facing right now. Like it it is unmanned systems that are being weaponized and some of them are being done in a kind of halfhazard way like IEDs were uh you know during during the the war on terror. Uh but you are

14:20also having very sophisticated state actors. The Chinese in particular have a variety of companies that are building weaponized drones with, you know, really lethal systems on them and those are being sold to other countries like right now.