In 5 to 10 Years, Using Weapons Without AI Will Be Considered Unethical | Yaroslav Azhnyuk, The Fourth Law

A Ukrainian entrepreneur who spent 14 years building cameras for pets pivoted to building cameras that down Shaheds, and is now building the autonomy software that could define how wars are fought for the next generation. Yaroslav Azhnyuk, co-founder of Fourth Law, joins Craig Smith in Kyiv to explain why Ukraine has become what he calls the Defense Valley or the Florence of Defense: a dense, fast-moving ecosystem of founders, engineers, and military operators who are building, testing, and iterating on autonomous drone systems in real combat conditions, with a feedback loop that no defense contractor in the West can currently match. The conversation covers Azhnyuk's five-level autonomy framework for drones, the eight-dimensional model for what a fully autonomous battlefield ecosystem requires, and the economic math that he believes makes global rearmament inevitable: a $500 drone that can already destroy a $5 million tank becomes roughly 10,000 times more capable when full autonomy is added for a few hundred dollars more. The competitive landscape is mapped with unusual candor, an "Apple vs. Android" comparison between Eric Schmidt's vertically integrated interceptors and Fourth Law's modular platform approach, alongside two arguments that cut against the mainstream narrative. First, that within 5 to 10 years it may become unethical to use weapons without AI, because non-AI weapons cause more collateral damage, not less. And second, that the real AGI risk isn't Skynet, it's the subtle transfer of power that happens when 100 smarter advisors gradually stop waiting for the President to decide, and nobody notices until the President is no longer the one making decisions. Subscribe to Eye on A. I. for weekly conversations with the people building and deploying the future of AI.

• 0:00Speaker 00

  There's lots of conversation about AI and defense. What we should be really talking about, I think, is software and defense. We started with making cameras for pets and now we're making cameras for drones. And we're throwing explosives to people who came to occupy our land. I think like within five to 10 years, it will become unethical to use weapons without AI. The same way as it might become unethical to use cars without AI. Because those are less lethal and non-AI weapons will be more likely to cause collateral damage, not less likely. But my only question to these companies is, if you're so great and powerful, why aren't your assets helping us fight off the Russians? Why aren't they on the front lines?

• 0:40Speaker 01

  Okay.

• 0:42Speaker 01

  So, so then you decided to get involved in, uh, in defense tech? And, and, and?

• 0:49Speaker 00

  Actually started a second company in defense as well, which, uh, makes, uh, cameras. Makes thermal cameras, which are now some of the most popular thermal cameras in the Ukraine military. They're used in interceptors. It was probably over a couple thousand chat heads down with drones with these cameras. And they're used in, in just frontline FPVs as well.

• 1:12Speaker 01

  Yeah.

• 1:12Speaker 00

  So it's, it's funny, like, there are some people from, from PyQ who also helped to start this company. And, uh, because they, they were doing cameras for the last, uh, 14 years.

• 1:23Speaker 01

  Yep.

• 1:24Speaker 00

  Cameras for pets.

• 1:26Speaker 01

  Oh, oh, that's right. Yeah.

• 1:28Speaker 00

  They have a lot of experience with that.

• 1:30Speaker 01

  Right.

• 1:31Speaker 00

  It's so ironic, like, yeah. Like, we started with making cameras for pets and now we're, we're making, uh,

• 1:40Speaker 01

  Cameras for drones.

• 1:41Speaker 00

  Cameras for, for drones. And we're throwing, throwing treats to dogs and now we're throwing explosives to people who came to occupy our land.

• 1:50Speaker 00

  So that's a pay flow of irony for sure.

• 1:53Speaker 01

  There seems to be a focus, you're, you're doing, uh, counter drone technology.

• 2:00Speaker 00

  Yeah, I mean, with, with the fourth law, um, we're, we're working on massively scalable, uh, autonomous systems for drones. And, um, at this point, it involves the, like, frontline FPV drones and then the interceptor drones, then some other drones.

• 2:20Speaker 01

  The interceptor drones, are those for FPV or are those for, uh, Shahed or what?

• 2:28Speaker 00

  Yeah, that's, that's a very good question. So, uh, currently like the most dynamic area that's had interceptors.

• 2:37Speaker 00

  But obviously, uh, when we're, you're talking about interceptor drones as class, that, that can apply to intercepting chat heads, ISR drones or reconnaissance drones. And then even smaller drones like FPV drones, uh, sort of frontline protection. So that's, it's a very important question.

• 2:57Speaker 01

  Schmidt's also working on?

• 2:59Speaker 00

  Yeah.

• 2:59Speaker 01

  Interceptors for Shahed. Is, is his, at Swift, is his company only looking at Shahed or are they doing the full range?

• 3:09Speaker 00

  Well, I think from, from the public sources, uh, they've been, uh, telling that, um,

• 3:18Speaker 00

  I think there are like at least three initiatives. One is Shahad interception, another is frontline autonomous drone. And the third one is something like mid-strike drone. So this, this strike drone capable of maybe, I don't know. I don't know what is theirs, but mid-strike as a category is up to 200,000 drones.

• 3:39Speaker 01

  Right. Yeah.

• 3:41Speaker 00

  And they're very capable, I think.

• 3:43Speaker 01

  Yeah, yeah, certainly. I mean, Eric Schmidt is no one more capable on AI.

• 3:49Speaker 00

  Like, they're our main competitor slash peer. So like, we don't strictly think of these companies as competitors because they're sort of trying to accomplish the same goal as we are. But, uh, yeah, obviously this episode in Silicon Valley, the show where they're like, uh, okay, these guys are making world better place, but we don't want them to make this better than we're making world better place. Something like that. It's not like that. So, uh, but there is, there is some front end competition there. But I, I think that are like, what, what they're doing is like for example, far ahead from what any other company, outside of Ukraine was able to demonstrate, far ahead of, say, Andrell or, uh, Alson.

• 4:29Speaker 01

  Right. And the, uh,

• 4:31Speaker 00

  I, I think we're, we're close to them in terms of, uh, the ground autonomy. Uh, it's like on, on grounds, so sort of FPV autonomy. And we're somewhat behind in terms of, uh, the aerial interception.

• 4:45Speaker 01

  Yeah.

• 4:45Speaker 00

  I think obviously they have a limited budget.

• 4:48Speaker 01

  Yeah.

• 4:49Speaker 00

  They have some of the best engineers in Silicon Valley, et cetera. And, uh, like, I, I want to try to do back of that.

• 4:59Speaker 00

  Hundreds of millions of dollars.

• 5:01Speaker 01

  And he has this incubator with D3, right?

• 5:04Speaker 00

  Yeah, it's like full disclosure, I was somewhat involved in, in starting that.

• 5:08Speaker 01

  In the, in the D3 incubator?

• 5:10Speaker 00

  A group of small, a small group of people who were sort of at the inception of that.

• 5:15Speaker 01

  Right, right. And is that funding primarily, uh, defense tech or are they funding everything?

• 5:23Speaker 00

  Yeah, I, I think their, their public stance is that they're funding, uh, defense tech. And, uh, but it doesn't have to be necessarily Ukrainian teams. I think part like, have about half of their portfolio is teams originally from Ukraine. Another half is, is not from Ukraine.

• 5:39Speaker 01

  Right.

• 5:40Speaker 00

  Because it's relatively small by, by Eric Schmidt's time. So I think it's like a couple dozen, uh, a couple dozen million dollars. This is like very small money.

• 5:50Speaker 01

  Is that how much?

• 5:51Speaker 00

  Maybe like, uh, it started with like 10, 10 million dollars.

• 5:55Speaker 01

  Oh, is that right?

• 5:56Speaker 00

  Maybe now like tens of millions.

• 5:58Speaker 01

  Yeah. And, and do you have any sense of their production versus your production? I mean, how many interceptors there?

• 6:08Speaker 00

  Nope. Oh, well, you, like, this is not something that you're supposed to sort of disclose to others. We do have sense of course.

• 6:18Speaker 01

  You, you have what?

• 6:19Speaker 00

  I think people in the industry do have sense.

• 6:22Speaker 01

  Have a sense. Yeah. I mean, but is he full production now? Are you full production? Are you still ramping up?

• 6:28Speaker 00

  Well, you're like, I think everyone is ramping up in this world.

• 6:32Speaker 01

  Yeah.

• 6:34Speaker 00

  It's like, obviously they're called Mero, the interceptors, uh, by Eric's company. I, I think they've started selling them to Poland and Romania. I think they had some actually like production roadblocks for Ukraine. Not sure, you know, what, what the plan is, uh, with their activity in Ukraine. Uh, there are also very capable interceptors by other companies, which are not as advanced with AI. But those, those companies intercept, I mean, they're buying AI modules from us.

• 7:12Speaker 01

  Oh.

• 7:12Speaker 00

  So it's a very different approach. I would say what Eric is doing is more a kind to sort of Apple strategy. They, they, they have the full vertical integration, they ship that to you. And what we are doing along with some other Ukrainian makers is more a kind to Android strategy, where different, different teams work in different modules. So for example, we provide autonomy modules, which, which are growing in their capacity. And those modules can be adopted onto various types of systems versus by, on the module from those.

• 7:50Speaker 01

  Because I was trying to understand where their models are coming from, where their chips are coming from.

• 7:54Speaker 00

  Yeah.

• 7:54Speaker 01

  He was saying they produce them in, in Ukraine.

• 7:59Speaker 01

  Is there a adequate semiconductor manufacturing or assembly in Ukraine to provide chips?

• 8:07Speaker 00

  Yeah, I think he might have misunderstood the question. Uh, there is no semiconductor manufacturing in Ukraine. Uh, maybe he was talking about the circuit boards.

• 8:18Speaker 01

  The assembly, yeah.

• 8:20Speaker 00

  Yeah. So, uh, you can, you can do sort of circuit board assembly in Ukraine.

• 8:28Speaker 00

  Yeah, there are some facilities where you can do up to eight layers. But I, I don't think they, uh, they are necessarily using them or, well, it doesn't matter. Uh, so you can do a lot in Ukraine. But then in terms of, uh, the actual chips or the sensors, like we're using, uh, sensors for thermal cameras, uh, they, these currently have to be bought from made and sent to semiconductor fabs. Uh, there are some plans to do, uh, semiconductor manufacturing in Ukraine, including our plan of building, um, uh, a thermal sensor factory.

• 9:09Speaker 00

  But that is still.

• 9:11Speaker 01

  They have this rocket shaped form factor that then flies like a cruise. Are you guys using something similar or are you using the traditional uh?

• 9:21Speaker 00

  We're like as as maker of autonomous systems, we're uh providing modules which can be integrated onto this far factor. So the most successful uh you can um intercepter is a company called Wild Cornets. And they're producing this drone called Sting. So it's a very similar form factor to general cherries. In fact, it was a some of those candle where uh Sting. Uh said the general cherry just copied what what they did. It's more or less like, you know, you made a car, so you copied my idea for a car. Um it's hard. It's an IP that's hard to protect. So we're working with all of these guys and there's.

• 10:03Speaker 00

  There are two major categories. One is sort of what we call a rocket drone or sort of a tail sitter drone with these. For um sort of wing propellers. Um another is a a more classic sort of plane form factor. Uh and there are a number of companies which are making planes.

• 10:23Speaker 01

  Um. Fixed wing, you mean?

• 10:25Speaker 00

  Yeah, yeah, yeah. Yeah, sorry. Yeah, fixed wing, fixed wing, um U U Vs. Um so we're working with with both of those.

• 10:33Speaker 01

  When you say autonomous, how autonomous are these?

• 10:36Speaker 00

  Yeah, yeah, like there's a rabbit hole here. So uh I sort of have this classification that uh. I suggested and uh most people in the market sort of accepted this something makes sense. So we talk about when we talk about the the individual autonomy for for single drone, we talk about this five levels of autonomy. And when we when we talk about the ecosystem, we talk about sort of eight dimensions of of of autonomous system. So the five levels for autonomy, sort of similar for five levels for self driving and you have. Level one is autonomous terminal guidance. So whenever you see the target as a pilot, you're flying as a pilot, non autonomously manually. Uh you target log.

• 11:19Speaker 00

  And after that, the drone is capable to autonomously hit a target and all the possible circumstances. That's.

• 11:27Speaker 00

  Level two autonomy is is sort of the same, but instead of hitting it, kind of as a style, you do bombing. So you you you calculate one to draw the ammunition and you do the escape.

• 11:38Speaker 01

  Correct.

• 11:39Speaker 00

  Um so that would be level two. Level three would be um when your system can. Identify the target honestly uh and make a decision to start. So it's it's flying. And then it scans whatever it sees through its camera. And then when it sees, it it knows that it is now in a mode where it's sort of in a kill zone, it's to find target. When it sees a target, it it first recognizes it and then gets enough confidence to make a decision, start strike. Which would be again, I think, level one or level two.

• 12:16Speaker 00

  Level four is navigation, so that you can navigate all the path from you start. Where the residue zone where you need to find that guy. Level five is is autonomous takeoff length. So you can do bombing, you can return.

• 12:30Speaker 00

  So that's for sort of for for flying drones, but then you can sort of adopt it to any kind of uh autonomous vehicle. Um and then when you when you look sort of further about the like for for us. Individual autonomy is is imperative, is is very important to accomplish and is very important to accomplish. In mass. So if the system that allows them to be in visual autonomy. Uh costs um even a couple of thousand dollars. It it it will be difficult to use such system on F P drones with just a hundred dollars per piece. But such system has to be relatively inexpensive and also relatively easy to manufacture so that it can be manufactured in millions.

• 13:14Speaker 00

  Because that's what the market demands. Um and then if you look further, uh how to the individual autonomy, you have sort of this. Autonomous autonomous battlefield. And there you have these. Eight dimensions of autonomous, which is high. So dimension one is levels of autonomy, five levels of particular destruct. Dimension two is the kind of carrier that you use. So it can be a propactor. Can be an F P V. It can be a larger sort of bomber. So it can be a fixed plane. Uh it can be a missile. It can be um even um shell, nuclear shell.

• 13:56Speaker 00

  And the uh man C vehicle or under C vehicle or man ground vehicle. Uh the third layer, uh the third dimension is dominant. So if you do. You you can you can use what culture, but it can be in. Ground to ground domain against enemy infantry or tank. It can be ground to air and interception. So it can be ground to sea, which is quite a different task. Uh and all the combinations of of of the of domains. So that's sort of the third dimension. The the fourth dimension is sort of what we call higher levels of autonomy. Things like swarms, drone carriers, drone nests, etc.

• 14:41Speaker 00

  Um then you have uh like number five would be different. Um outside conditions. So day, night, uh winter, summer, um different types of targets. Uh because attacking an infantry unit or attacking um. A tank or or or vehicle are slightly different tasks which which requires some different developments. Um then you have uh level six, you have a command and control uh systems. So. As as commander of a brigade, she's like two to five thousand people or. The ports can be up to forty thousand people. Um.

• 15:25Speaker 00

  How do you manage all of your units with the uh these autonomous assets? How do you manage all the data coming from those assets? How do you command, analyze, etc? Um then on level seven, you have all the infrastructure needed for all this to work. Uh starting with uh simulations, which are used routinely for testing, etc. To um. Things like uh data collection mechanisms, security subsystems, um remote deployments of systems. Um and then finally, uh sort of dimension number eight is. How massively deployed that system is? Because if you have one system.

• 16:09Speaker 00

  Uh deployed in hundreds of thousands of units and another. Just passed the field test provisions and are just being tested, you know, a couple hundred units being tested. Those are very different systems, very different levels of capability. Uh so that's sort of how we think about this whole ecosystem. And I think it's like what's going on here is just astounding. This is. This is like the defense valley.

• 16:34Speaker 01

  Right, exactly.

• 16:35Speaker 00

  Like you're now in this room, there is a guy who. Like sits somewhere here, who works in the industry, who I know very well. And this happens to me all the time. Like I I come to meet some defense technology founder. And then there is another defense technology founder working today. And we introduce each other. So it's a very, very like Silicon Valley uh vibe.

• 17:00Speaker 01

  Yeah.

• 17:01Speaker 00

  And lots of confidence demands. The ideas flow.

• 17:04Speaker 01

  Yeah.

• 17:05Speaker 00

  Lots of expertise. So even with the domain expertise, no. I say this all the time, that my definition of strategy is that. Strategy is a plan of victory using the uniqueness of your situation. The uniqueness of Ukraine's situation is that we've been at war with Russia for twelve years. And no one else in the world wants to steal that advantage from us. So you have a couple million people. Uh who were one way or another involved in this. So either as fighters or as makers, engineers, entrepreneurs.

• 17:42Speaker 01

  Right.

• 17:43Speaker 00

  Or as assembly workers. Hmm or as volunteers, operation specialists, governmental clerks. Hmm. So this is an incredible domain experience and it will remain so. For decades to come, even if tomorrow there is ceasefire. And um I like, for me, it's obvious the world will be rearming for the next couple of decades. Uh driven by um supply chain instability, by. Obviously advances in drones and robotics and obviously by A I. So three huge factors that will drive the global rearmament. Conventionally and unconventionally, I think we have probably.

• 18:28Speaker 00

  Another dozen or dozen and a half countries in the world who are now working on their nukes. Who do not have an official nuclear status. So unconventionally, I think they'll they'll also go to the right.

• 18:41Speaker 00

  Again, because of sort of Russian invasion of Ukraine and the special memorandum and sort of the current policy, the current policy by the US, now there's this very unique moment in time for Ukraine, which can be a cornerstone of not just Ukrainian but European strategy, building, you know, the Florence of the fence here. And this is happening, I think the government can definitely accelerate that if they pass some smart regulations. But one way or another, Ukraine and Ukrainians will play a critical role in the rearmament of the West.

• 19:22Speaker 01

  Do you think that the industry will turn back to commercial consumer tech after the war? Do you think that not the defense of a defense industry has been built, but Ukraine will become a supplier?

• 19:37Speaker 00

  Absolutely, but I think that when I hear after the war, I imagine something quite different. If we have ceasefire with Russia in 2026, this is not after the war. After the war is when the global arms race is to a sense still. Before that, unfortunately, in like the next couple decades, the whole world will be rearming. And now, one piece prepared for the war. So that's what everyone is going to be doing because these systems, they just change the equation so much, like can take out five million dollar tank, you know, the drone costs them five hundred dollars.

• 20:19Speaker 01

  Yeah.

• 20:20Speaker 00

  And if you can add autonomy to that drone for the cost of a couple hundred dollars on top. And, you know, and autonomy itself makes every drone usable by a hundred times as many people, if you don't need to train as pilots, makes ten times more precise, makes it not one time, but multiple time bombers, probably ten times more utility. So you guys, this be order of 10,000 times improvement by adding complete autonomy to a system. Five million dollars and five hundred dollars, how many orders of magnitude is that? It's like thousands? 10,000. So it's 10 to the, like, 10 to the fourth, multiplied by 10 to the fourth, it's like

• 21:05Speaker 00

  10 to the eighth. It's like, what, 10 million, 10 million times, 100 million times, 100 million times more people than current systems. It's like, it's hard for him to imagine what happens. And then, like, if we talk about threats, like, in, in September, there were 20 Russian Shahed inventors, Herbera, or what they call it, flying over Poland.

• 21:46Speaker 00

  Only four of those were shut down with equipment that cost probably two orders of magnitude more than, than the assets. General Secretary Rutte had a press approach, saying, you did a fantastic job, we've shown resolve of NATO, how we can protect our land, but it's obvious to everyone, it's a huge disaster.

• 22:11Speaker 01

  Yeah.

• 22:12Speaker 00

  Now imagine Russia attacking with, but those were even drones without any explosives in them. Imagine Russia attacking with 50 of actual Shaheds.

• 22:21Speaker 01

  With 50?

• 22:22Speaker 00

  Shaheds. So they're attacking an airport in Jashu, a city where close to Ukraine, lots of Ukrainian equipment is being pumped there.

• 22:31Speaker 01

  Well, that's, you know, when you're talking-

• 22:33Speaker 00

  Russians say, you know, it's your legitimate target. Attack that. What will Europe and NATO do? Will we send 50 Storm Shadows to some obscure Russian city? Well, most people I talk to in the EU say that we won't have the guts to do that. But even if, will it be sort of an eye for an eye? No, it'll be huge defeat for Europe, huge defeat for the free world, it will be a precedent, and Russia doesn't care, Ukraine blows your cities every night, so what?

• 23:10Speaker 01

  Yeah.

• 23:11Speaker 00

  And obviously, huge impact on the markets, huge impact on the economy, huge impact on everything. Now, the second scenario is Ukraine produced about 4 million of these inexpensive FPD drones last year, 500 dollars each. So if Ukraine can do that, by the way, United States cannot do it today. You cannot, you can just, there's no capacity to produce that many drones here. But if Ukraine can do 4 million, China can easily do 4 billion.

• 23:45Speaker 01

  Yeah.

• 23:46Speaker 00

  And then, if they really want to, they can probably do them, not with propellers, but with wings, so fixed wing, which, which is launched from a tube, which flies not 30 kilometers, but maybe 200 kilometers.

• 24:03Speaker 01

  Yeah.

• 24:04Speaker 00

  And DJI is the world's leading drone technology company, and I'm sure they have the autonomous capabilities they need. And then, Chinese shipbuilding is also very impressive. So can they build thousands of autonomous submarines capable of carrying those drones? I'm pretty sure they can. So what if China builds 5,000 underwater submarines, carrying a couple hundred of these autonomous drones, very inexpensive, capable of going to 200 kilometers inland

• 24:42Speaker 00

  each, and shows up to the coast of any coast, of California, or Taiwan, or Great Britain? Is there a good response to that currently? I don't think so. Hunter subs won't be enough? Hmm, well, like, that's, that's a very interesting question.

• 25:05Speaker 01

  Yeah. One of my questions that I've been asking, why hasn't that already happened? Because Russia does not have very strong AI research, but this is not cutting-edge research, this is pretty much a solved problem.

• 25:26Speaker 00

  Well, that's what I thought.

• 25:29Speaker 01

  And you would think that-

• 25:33Speaker 00

  But it's not about only Russia, like even the United States doesn't have the capability.

• 25:36Speaker 01

  No.

• 25:37Speaker 00

  The United States doesn't have autonomous drones, which don't rely on GPS, and it doesn't have them, that's an inexpensive cost. And what it has, well, I'm sorry to say this, but I'm not sure if it works, because I have tremendous respect for companies like Andrell.

• 25:59Speaker 01

  Yeah.

• 26:00Speaker 00

  But my only question to these companies is, if you're so great and powerful, why aren't your asses helping us fight off the Russians? Why aren't they on the front lines?

• 26:16Speaker 01

  Although, this is what interests me about Eric's-

• 26:21Speaker 00

  He is different in that extent, because his stuff actually works.

• 26:24Speaker 01

  Yeah. And, and my guess is that for him, Ukraine, as you said, there's going to be a rearmament. Ukraine is like a, you know, a development zone and field testing.

• 26:49Speaker 00

  Yeah. Yeah, look, I think Eric is a remarkably smart guy. I had a chance to meet him a couple times. Hmm, and he just saw around the corner before everyone else. Remember, he was doing work for the US Department of Defense before Russian invasion of Ukraine.

• 27:10Speaker 01

  Yeah.

• 27:11Speaker 00

  And he also had-

• 27:15Speaker 00

  The luxury of a lot more free time than many others.

• 27:20Speaker 01

  Yes.

• 27:21Speaker 00

  He was able to think deeply and think ahead about these problems. And that's why he focused so much of his attention on the plan. And still is focusing. Um, I, I think like one of the smartest things that he did is actually hiring a lot of people here. And very quickly iterating with, uh, with actual military units. Uh, because you, you cannot, you know, it's like. I, I went through YCOM nature program in 2016 and, and then. The one thing they, they, they teach, uh, folks who get in is. I like actually they teach everyone, talk to your customers. Talk to your customers. This is the most important thing that you can do as, as a company. And, um.

• 28:04Speaker 00

  Well, obviously for Andro, the customer is the bar of the fence. They're, they're doing their job. For HMed, the customer is some future fighters, which will fight future wars. And so because the future is currently here.

• 28:19Speaker 01

  Yeah.

• 28:19Speaker 00

  For, for us here, he comes here regularly. He has a white team here for deployed engineers working with particular brigades. And that's why they were able to build something that actually works. And that, that actually saves lives because they, you know, their essence intercepts. Chahads above our heads and, um, you know, I'll save lives of infantry by, uh, now destroying Russian advances. Um, so I think that's why he's succeeded. And I think other companies are slowly coming around to that. And, um, there's more and more, um, for foreign officers talent, uh, coming in here. And that's why, that's why it is the, the, the defense valley, the, the, the Florence of defense.

• 29:05Speaker 01

  Yeah. Why do you think he's avoiding press?

• 29:09Speaker 00

  Oh, I think it's. It's funny, um. Because. He's been avoiding press, but then from time to time, there are these huge press releases, which is like. This is so inconsistent with that strategy. Like the first one was when, I think the last year in Denmark, they signed like a big deal with Zlansky. And they were like photos of Eric Zlansky.

• 29:35Speaker 00

  Um.

• 29:37Speaker 00

  And I was like out of the blue. I saw that. It was like, okay, strategy change. And then again, like they, they, they went back into silence and then. I think there was news of, uh, them doing a deal with Poland. Uh, so that was again sort of public. So I have no idea. Um, I, I, I think it's just, uh.

• 29:58Speaker 00

  I think there was a good team of lawyers for thinking this through.

• 30:02Speaker 01

  Yeah. Yeah. Uh, on autonomy, I'm trying to understand these levels of autonomy. You know, I, I worked on the National Security Commission on AI, which Eric. Uh, was the chairman and, uh, was a Pentagon funded. I think you were referring to his, uh. Work with the, uh, I've forgotten the name, but the, the outfit in California that was, uh.

• 30:31Speaker 00

  I think he was like, like either an advisor to the government, something like that on, on, on the matters of the defense technologies.

• 30:39Speaker 01

  Right. Yeah.

• 30:40Speaker 00

  Even, even like writing about that along with some, uh, accomplished, uh, you know, people in the military. Uh, like you can tell he's been thinking deeply about these problems.

• 30:53Speaker 01

  But they make it very clear when they speak in the United States, that full autonomy. Uh, is off the table. The, you know, legal autonomous lethal autonomous weapons, uh, that, that target, uh. Autonomous lead, uh, is something that they will not do.

• 31:17Speaker 00

  Hmm. Well, what if China does it against?

• 31:19Speaker 01

  Well, that's right. I mean, that's what I say. That's, that's what they say. But everyone knows it's like Iran, Iran, uh.

• 31:27Speaker 00

  Nuclear.

• 31:28Speaker 01

  With, with their, uh, enriching uranium. They say, oh, we're only going to enrich this much for fuel, but to get to the next level, you just run it longer. So.

• 31:37Speaker 00

  Yeah. You just flip the switch basically. So, um, and then, um, you know, it's, uh. I think that the, this is made more complicated by the press than it should be. The, the ethics of AI in defense. Let's take an example of an artillery shot. When do you make a decision to, uh, strike a target with an artillery shot? Well, in the moment, well, when you pull the trigger and you fire the gun. So you cannot pop an artillery shell 500 meters away from the target and, and. Uh, ask for a human confirmation, uh, whether it's the right target or whether we, uh, shot, you know, two kilometers in, in wrong direction and, and hit the village.

• 32:20Speaker 00

  You can think of an autonomous drone as an artillery shell with propellers. And you make a decision the moment you launch, the moment you give command. Uh, so that makes things much, much easier. Uh, the second thing I want to say on this topic is that. I, I think like within five to 10 years, it will become unethical to use weapons without AI. The same way as it might become unethical to use cars without AI. Because those, those are vastly thrown and n- n- non-AI weapons will be more likely to prosper. It's not less likely.

• 32:57Speaker 00

  And the third one is you can be the good guys, but bad guys are not going to abide by the rules.

• 33:04Speaker 01

  Right.

• 33:04Speaker 00

  So. We, we gotta have at least the capability to, uh, have these systems working not at the speed of. A human command chain, but the speed of, of the software branch. And also, you know, there's, there's lots of conversation about AI in defense, but, uh. AI is just sort of the latest big thing. Uh, what we should be really talking about, I think, is software in defense. Uh, because a lot of these new capabilities are created not with AI, just classical software and. Like even AI software, right? Um, and then, and then like, look, even AI is such a buzzword, right?

• 33:49Speaker 00

  Alex, Alex Matt is what 15 years old. Uh, and like image recognition capabilities are, you know, at least a bit kated and they're more in. Sort of cases and a lot of what is called AI for defense is just image recognition. Um, and then obviously we're interacting with large language models now as consumers. So people think that, okay, a drone with AI, something like a chat GPT with propellers that can explode. It's not like that at all. It's far from that. And, uh. If you look, if you think about the robotic system, in an economical sense, you have sensors in.

• 34:33Speaker 00

  Uh, so you have basically physical, physical data receptive by hardware, by sensors. Then you have three software layers. You have perception, planning and control.

• 34:43Speaker 01

  Right.

• 34:43Speaker 00

  And then you have actuators out, so commands to monitors which move the thing somewhere. Um, or make things go, uh. And those layers, perception, planning, control, each of them can be done by classical software, lots of expertise. Sure. Literacy on that, lots of evidence, lots of robotic systems which do that. And then each of them can be done by neural net or all three of them can be combined into one neural net. Or they can be grouped two by neural net, one by past problems. Um. So we're, we're definitely sort of going towards the world where these systems will be able to do.

• 35:24Speaker 00

  Control on their own and some will have some freedom in decision making, whether, you know. But the first decision these systems are going to make is, okay. Enemy vehicle is hiding behind a building, whether I go around the building from the right side or from the left side.

• 35:42Speaker 01

  Yeah.

• 35:43Speaker 00

  Um, and for years ahead, uh. Uh, and AI enabled defense system will be no closer to a Skynet appraising that. Than, than a warehouse of 155 caliber shells. It, it's just, it's just not there. Uh, and then I think before we are at a risk of being wiped out by, uh, defense Skynet. We will, uh, uh, be many years under risk of being wiped out by, um.

• 36:23Speaker 00

  Advanced AI, uh, developed on, on servers of one of these big hyperscale.

• 36:29Speaker 01

  Yeah.

• 36:30Speaker 00

  Um, so in that extent, I think, you know, we, we should be careful about AI. But first and foremost about, uh, sort of AGI and even with AGI, I'm not. Um, I'm not as afraid of it in the coming, uh, years.

• 36:45Speaker 01

  So I think that that's, it's more subtle than, than Skynet or, you know.

• 36:52Speaker 00

  Yeah, that's, I think that's, that's a good one. Um, and then also sort of raises, uh, goes back to the question of sovereignty. And, uh, like people mostly talk about the sovereign supply chains. But there is also sovereign technology and sovereign know-how. I think it was yesterday was in the news that, uh, Donald Trump said that they, they might, uh. Ban Chagaki in Germany or something like that. And that, that basically sends the signal to not just to Germany, but to entire world. You've got to develop your own foundation models. And obviously one person that's very happy is Yen Sun Huang. Uh, because, uh, he's like, that's, that's his method box. Like every country should have their own sovereign and like, of course they, they should buy our chips.

• 37:38Speaker 00

  But he's, he's a very good business.

• 37:41Speaker 00

  Uh, one, one thing I think about the AGI is actually that. I think there's very high probability that, um.

• 37:51Speaker 00

  To develop AGI, it will have to

• 37:55Speaker 00

  it will have to get into physical bodies, it will have to be

• 38:01Speaker 00

  like, these agents will have to be actual robots.

• 38:06Speaker 00

  Because I think the good analogy

• 38:10Speaker 00

  I sort of came up with is

• 38:13Speaker 00

  it's like the Plato's cave, like the current L M's are living in a Plato's cave and they're learning about the world from the shadows on the walls, which are what humans wrote about the world. But they cannot experience the world themselves.

• 38:26Speaker 01

  Yeah.

• 38:26Speaker 00

  So for them to be able to experience the world, they would have to have the bodies and walk out of the cave. And I think it's also like a good number of researchers who are sort of working in that direction, like T F A Lee, I think is like her work is very impressive.

• 38:44Speaker 00

  And yeah, I think like some people in the industry share this intuition as well.

• 38:49Speaker 01

  On the autonomy, when you're saying the decision is is whether or not to launch, but once it's launched, navigation, targeting, and execution or are all autonomous, that has to be on board, that model. You can't be connecting to the cloud.

• 39:13Speaker 00

  You can have you can have connectivity as an option, right? But you have to be able to have a fail-safe mechanism, like what there is no connectivity can be jammed. The the point where it connects to can be destroyed.

• 39:33Speaker 00

  So you've got to, you've got to do all of that on the edge. And actually, so you said it starts when it's launched, but it will gradually start even earlier and earlier. So imagine autonomous system detecting and then launching. Right, then autonomous system launching systems which were detected. So it's like this, this Victoria computer game. So you just keep automating, automating, automating, higher, higher.

• 40:05Speaker 01

  And and the drones that you're building, how autonomous are they at this point?

• 40:11Speaker 00

  So we are building drones, but we also building the autonomy modules, as we call them. And sometimes we just even license the manufacturing of these modules, license them out. And we license out the software. So we then we just produce the software which the manufacturer of drones installs in their drones. But we do make our own drones as well. It's just like making drones is sort of a more of a commodity. And the really like interesting thing about this is you know, drone is like hamburger. So you can make a hamburger in a kitchen, but the really difficult thing is making them announce.

• 40:49Speaker 00

  There's a may make a factory which makes same hamburgers with a very short timeframe, can make millions of them without onions or with onions. You know, and that's what the Ukraine drone making companies are. But what we ship currently in mass is sort of level one autonomy for ground interception and we call the level zero for aerial interception, which is a detector, is the AI detector, which helps pilot to see the target, but doesn't really take over the control of it now. And the other systems we have on various levels of development and deployment.

• 41:32Speaker 00

  There are five systems which are sort of ready to be fielded and are in the process of early testing by the national.

• 41:45Speaker 00

  And yeah, it is like, as I've said, there are so many like different dimensions to this, right? So there's like same level of capability, but on different platforms or in different domains. Or, you know, so daytime, nighttime camera, etc. And obviously, you know, not everything I can talk about.

• 42:07Speaker 00

  But yeah, it's a, it's a very exciting and fast pace. And then remember,

• 42:14Speaker 00

  in 2017, Vladimir Putin said that ever will master AI will rule over the world.

• 42:25Speaker 01

  I remember.

• 42:26Speaker 00

  Like, just think about how evil that phrase is, will rule over the world, will not cure all the diseases, will not save the humanity. It's like, it could have been like, even though in Putin, it could have been save the humanity, right? With his K G D lies, but no, he was very explicit what he wants to do.

• 42:48Speaker 01

  That's right. Although although he hasn't done a very good job of of mastering AI.

• 42:55Speaker 00

  Well, that's just classical Russia. They they lost everything they could. They lost their their space technology and like in in in 2011, when the US sunsetted the shuttle program.

• 43:11Speaker 00

  So you was operated by Russia, but developed by Ukrainian Korlov, because that's Korlov's legacy. Korlov is Ukrainian. He basically fought of himself as Ukraine. The first first ever song which was sang in space to Korlov's

• 43:27Speaker 00

  wish was a Ukrainian song, which literally says, why am I not a falcon? Why don't I fly?

• 43:36Speaker 00

  So basically, when when when when the American program wasn't ready, you know, everyone flew into space on a Ukrainian ship operated by Russians. And then now they have their planes falling apart. And that they were not able to build their their own plane. They have huge setbacks in their rocket programs. So it's it's a country that's completely destroyed by corruption and incompetence, but still can do a lot of harm.

• 44:06Speaker 01

  I'm trying to like zero in on autonomy because some people say, well, the AI isn't there yet. We don't have models that can target precisely, for example, cannot differentiate between a tank and a civilian car. I don't think that's true.

• 44:25Speaker 00

  Yeah, like, come on, like, who believes that? Of course we can.

• 44:30Speaker 01

  So why are

• 44:34Speaker 01

  the drones produced by Ukraine, why aren't they given that level of autonomy? Is it a cost issue? That that or a supply chain?

• 44:45Speaker 00

  I'll ask you a couple of other questions in that, in that line. Why aren't drones produced by the US don't have that capability?

• 44:51Speaker 01

  That's right, but but the US is not in an active.

• 44:54Speaker 00

  Yes, then why don't we have self-driving cars?

• 44:58Speaker 01

  And but that's a regulatory issue. It's not.

• 45:01Speaker 00

  Not really. Like, the technology isn't ready. And the reason and and.

• 45:05Speaker 01

  It's it's close. I mean, way more operates. It's going to be.

• 45:09Speaker 00

  I know, I know, I know, but it's been close for what, 10 years more? Well, I mean, Sebastian from started in 2007 and NASA challenge was even before that.

• 45:22Speaker 00

  The the answer, and then also, why, why don't we have AI agents all over the internet doing like useful work? Why didn't they replace you and me and all of our friends? I think the answer to all of those questions about why isn't AI more widespread is the edge cases. Like thousands and millions of edge cases which have to be polished out. And the thing with the warfare is that if you have

• 45:52Speaker 00

  say an F P D drone is a system that in, you know, used by millions, but what is, what is its effectiveness? Well, it's between 10 and 20%. 10, 20% of these drones which are used actually could turn and explode and cause the desired.

• 46:08Speaker 01

  You're talking about the interceptor drones?

• 46:10Speaker 00

  No, I talk about the frontline.

• 46:12Speaker 01

  Right.

• 46:13Speaker 00

  So that's a low sort of mark to overcome, right? And that's why drones like ours are constantly used in the frontline and they increase the efficiency by two to four times, even with level one autonomous system. And then as you go further with wider, wider autonomy, you basically you multiply edge cases. And then add to that, not autonomous, but robotics problems. Like, most of the things we deal with is not about AI, is is about software and robotics.

• 46:51Speaker 00

  So from the outside, because AI is such a buzzword, it might seem that a drone company that does AI has, you know, 80% of its staff or engineering staff doing AI. But it's more like 10%, you know, because there are so many other problems.

• 47:11Speaker 00

  So it will happen. It just, it will, it will just have to go through all of these issues. And it's hard to predict when, this is always this this ask curve, right? You go over some threshold and then you have this take off.

• 47:25Speaker 00

  I certainly predict that it's gonna not take as long as it took with self-driving cars.

• 47:30Speaker 01

  Yeah, and and regarding the US, I mean.

• 47:35Speaker 01

  There's a existential imperative in Ukraine that so things are moving much faster than elsewhere in the world. Do you in terms of the target navigation has been pretty much solved, right?

• 47:55Speaker 00

  Not really, unfortunately, yeah. It's like, again, this is like, there's so much sort of coming, like, you would, you would think it was solved, right? It's not, because if you think about the navigation problem, well, first of all, it's not one problem. It's like a hundred different problems, because navigation for a $500 drone, which has to fly 40 kilometers and maybe 20 kilometers, and then return back and flies at speeds of up to 100 kilometers per hour, or 60 miles per hour. That's one problem. Navigation for a wing missile that goes slightly subsonic is another problem.

• 48:33Speaker 00

  Navigation for, you know, I was like, last night, well, corrections for hypersonic is, is a very different problem as well. You're flying above, uh, the altitude of flight is very much, um, important for navigation. If you're flying high, it's easier to navigate. Flying low, you have less data about what's, what's meant to you. Uh, then you fly, uh, above a body of water, much more difficult to navigate than a body of land, uh, if you fly at low altitude. So all of these factors, the length, the cost, the altitude, uh, the terrain, all of them impact the navigation system, and there won't be like one AI, you know, Google Maps that solves it for anything.

• 49:15Speaker 00

  And customer phone is on their smartphone. Um, it's going to be very different than that, although maybe at some point there will be, uh, something like, like I just described. Um, and then, but yeah, you were about to ask about the, the target detection, actually.

• 49:31Speaker 01

  Yeah, so the target, that part is solved, right? But you have a camera, you have, uh, object recognition, either thermal or, or, uh,

• 49:43Speaker 00

  Yeah.

• 49:43Speaker 01

  segmented, uh, and you, the, the drone hits that target, I'm talking about kamikaze drones, right? Hits that target.

• 49:52Speaker 00

  Yeah.

• 49:53Speaker 01

  Why isn't, or is that already being used?

• 49:58Speaker 00

  Well, it is, it is solved to some extent, right? It's, it's easy to spot, um, you know, a green tank on, on, on, on a, a white snowy field. It's much more difficult to spot something camouflaged than, uh, you see Russians riding on donkeys. Uh, like, uh, was that in the data set? Probably not. Um, then you see Russian, the Russians dressed up like penguins. Was that in the data set? Also not really. Um, then, um, you know, when you talk about the practicality of this, um,

• 50:40Speaker 00

  it's not just about the algorithm that can recognize an image. Uh, so you have an MFP drone that's autonomous. How fly, how, how I should it fly? If it flies lower, it, it's closer to what it has recognized. It can maybe recognize it, um, better, um, but it, it has a smaller field of view. And how fast does it fly? Does, does it scout the territory or find something? Or, or how does the recognition, uh, integrate with navigation? And then how does that integrate with the tactics of its usage? Like, do we need autonomous drones against enemy infantry? Or do we need against enemy logistics?

• 51:22Speaker 00

  Um, there are so many of these questions, um, that in, in practice, uh, going from the technology of being able to recognize something to the device that's actually usable, useful, usable, and creates more utility than the existing alternatives, is not a super obvious thing.

• 51:44Speaker 01

  Yeah. Are you worried about fully autonomous, just in terms of the, the human cost of war?

• 51:54Speaker 00

  I'm not sure what you mean by that question.

• 51:56Speaker 01

  If you have fully autonomous AI-guided fire-and-forget drones, maybe even launched.

• 52:07Speaker 00

  Yeah.

• 52:08Speaker 01

  Uh, that's a pretty terrifying prospect on both sides.

• 52:12Speaker 00

  Oh, no, like, it's terrifying if the enemy has it. If we have it, it's, it's a savior.

• 52:17Speaker 01

  That's right.

• 52:18Speaker 00

  And we, if, if we don't have it, that is what is terrifying, right?

• 52:22Speaker 01

  Yeah.

• 52:23Speaker 00

  And like, you, obviously, you need to have autonomous interceptors to counteract enemy autonomous drones, because, uh, they mass and, and like, like how many of them they can deploy will, will not be as limited by humans now. And then obviously, even now, it's hard for human to match the speed, agility of machines. That will be, become harder and harder. Um, so I am, what I'm terrified about is that we won't get, that we won't get there as fast as the enemy does.

• 52:59Speaker 01

  Yeah.

• 52:59Speaker 00

  That's I'm terrified about.

• 53:01Speaker 01

  Yeah. Okay.