Angry Planet - Catching Up With Killer Robots

Episode Date: May 22, 2018

When the United States put a machine gun on top of what looked like a bomb-squad robot, it didn’t care much who it was aiming at. But on the whole, drones and other killer robots are pretty effective, and there’s more to come. This week we speak with Kelsey Atherton, a writer for C4ISRNet and Fifth Domain, who brings us up to speed on these deadly machines.Support this show http://supporter.acast.com/warcollege. Hosted on Acast. See acast.com/privacy for more information.

Love this podcast. Support this show through the ACAST supporter feature. It's up to you how much you give, and there's no regular commitment. Just click the link in the show description to support now. To the sense that there is any vague notion of fiscal responsibility and defense planning, sending a robot to follow a submarine for a couple weeks is way cheaper than sending a destroyer with a couple hundred people on board to do the same. You're listening to War College, a weekly podcast that brings you the stories from behind the front lines. Here are your hosts, Matthew Galt and Jason Fields. Hello, welcome to War College. I'm Matthew Galt. And I'm Jason Fields. The U.S. Air Force has retired the MQ1 Predator Dron, the iconic hardware that's come to define the back half of America's wars abroad.

So what's next for the future of autonomous warfare? Kelsey Atherton is here to help us sort it out. Atherton is a staff writer at C4 ISRNet and the editor-in-chief of Grand Blog Tarkin. His work has appeared in Popular Science, Aviation Week, and Vice. Kelsey, thank you so much for joining us. Happy to be here. All right, so why is the predator retiring? It's not that old, is it?

It's that old. We don't think of it as having done this, but it flew, I believe it flew in Kosovo. So that would be the 90s? That would be the 90s. Yeah, and the designs go back further. We're flying older planes, and we're flying A10s. Most every fighter we have that isn't stealthy in some ways significantly older, but those aren't essentially lawnmower engines, pusher props on just matchstick body wings.

It's an older machine. It's iconic, but it's old. All right, well, what about the Reaper? Does that still have some life in it? The Reaper absolutely has life in it. Technically, the Predator B, so if we want to see the predator as continuing on, the Reaper is also there, since it's confusingly also named that. It definitely has years in advance.

In the Pentagon's budget request, I believe they included maintenance and, I think, repair parts for it. I don't know if they're getting new, but the fleet that they have, they're pretty reliant on. They're pretty happy with the drones they have, and they want to keep using them as they've done so. Can I ask what the difference is between the two? Sure. So the Reaper can carry more and fly longer. It's essentially just a very upgraded version. The Predator was originally built exclusively as a surveillance platform.

And then they added on the spots to put two missiles on it. The Reaper can carry, I believe four on each wing. And it can fly for up to, I think I saw 27 hours is the number I saw most recently. that's endurance limits, but essentially the Reaper is the predator, but more so, the next generation and better and bigger in every way. All right, so you called them lawnmower engines on prop planes, basically?

Well, what's next? What's the next advancement then? So it's hard to say. It's not going to follow the traditional path that manned aviation has so much. We already have some drones with jets. The global hawk is the logical end sort of of this existing body type. The global lock has more sensors, flies higher,

has jet engines. I think it can be airborne for 30 hours. So that's one thing, but that already exists. And there's only so much you want to do with that because it's a lot of work. These are also, I should be very clear, these are remotely piloted vehicles. The Air Force likes to use remotely piloted aircraft or remotely piloted system as the term, which means that there's human pilots constantly in charge of them. They aren't so much labor-saving as they are, the fact that the pilot can, like, go take a nap in the middle of the flight

and someone else can start flying it. But there's always people actively controlling it. So if we're looking for what comes next, if we're looking for the way drones go in the future, it's less about, like, oh, well, it'll be faster or bigger or necessarily fancier. There will be probably some of those features. There will probably be some stealthier bodies.

but the big change isn't about the airframe, but about the software. And that's going to be in sensors. It's going to be in artificial intelligence. It's going to be swarming, possibly. There's a whole list of functions, which is essentially taking advantage of how sensors and computer controls work to make it to that the drones need just human supervision more than human direct control. Swarming? Swarming. Sounds cool. What's that?

Swarming is cool. Swarming is, and we see this on smaller drones. We have yet to see it in anything super big, though it's possible. But swarming is where many drones flying all communicate with each other and give each other information about where they are, and then they can work together as a swarm, and we think of swarms as this big, uncoordinated mass that goes from one place to another, and it looks very messy.

But if you see the light shows, and there's a couple companies that really, really, love to do these big light shows. I know Intel keeps setting records that it can then incrementally break over and over, but they do light shows with drones where they have small, mostly quadcopters. They have an indoor version and an outdoor version, but these are quadcopters with LEDs. And there's a ground station that relays controls to the swarm. But broadly, the swarm operates where the drones will fly and they'll do like a swirl of light, so they'll do big shapes. I think there was a demonstration as part of the Olympics opening ceremony. And that's what,

like, that's the art and the civilian side of things. And the military utility of something like this is unlike predators, reapers, and global hawks, which are pretty expensive, if you have a lot of relatively cheap drones, and some of these we're talking hundreds and even if we're talking more than that, it's low thousands as opposed to many millions. If you have a swarm and you give it a task, right? Swarm, go scout, this area, go scout over this hill, go check out that mountain, go bug these guys by that missile installation. If some of them get shot down, the nature of the swarm means that the whole swarm can still do the job. They will communicate with each other. They will fly there. They will record what they need to. They will transmit back what they can. And you'll have to get all of them

to get the sky rather than just shooting down like one plane and then there's nothing, no surveillance. You have to shoot down a lot, a lot of drones in the swarm for the swarm to not work. What about some of these counter drone measures that we're seeing? You know, you see these giant kind of drone cannons, anti-drone cannons. Do you know I'm talking about? I think so. Are you talking about like the ones that are like big fancy antennas? Are you talking about something else?

The big fancy antennas. Yeah. So that is a big part of it. How to stop a drone is a growing market on the center for the study of the drone at Bard University. They do great research on this. I crib whatever I can from them. They had a recent report on counter drone market, and this is looking at the systems that, like,

you can get brochures for. If there's something that's classified of existence, and we don't know that. But they looked at, I think, 200 and I want to say 34 systems that are counter drone. And the overwhelming majority of systems that can stop a drone or do something to mitigate the drone flying are directed antennas that either spoof or jam or.

or sometimes feed it code. And that's just because the nature of the drone is that a lot of them, especially the commercial ones, they're over their basic, like, Wi-Fi frequencies or their cell phone frequencies. And so it's not super hard to get in there. I once saw a demonstration, actually, at West Point, where they had a, God, I think it was $50 all put together between the antenna, the, like, Airsoft Rifle Butt and the Raspberry Pi,

and they made a what they called a cyber rifle. And they demonstrated it where the attacking squad had a, had a cyber rifle specialist with them, and the defending squad had a quadcopter to, like, try and scout the area. And so they were, um, this was, this was like June of 2016. They were experimenting with what is it like to be in combat, what are the ways to get it down. And this was like, it was a parrot quadcopter. It was pretty simple. And the cyber rifle in question only really, like, worked on that model of parrot.

This was very demonstration. But there are other ones, and we've seen them deployed. In Iraq, we've seen tested and purchased by the U.S. government and likely by others. And they are designed to find what frequency the drone is on and either give it the order to return home or land. If they can't give it a order over that, they will overwhelm the processing and make it just basically shut down where it is. And there's an offense defense dynamic to this as there always is.

I know Patel, who makes the drone defender that we saw in the rock, they had to update a version because ISIS figured out what frequency their drones were being jammed on and started using different frequencies, and the original rifle couldn't quite get there. But that's this current state of play, really, of character drone stuff. Do you ever see anything like that working on something bigger, a predator or a reaper? Absolutely not. You would have to, the range on these is just incredibly limited. I don't know if I've seen one that has, that boasts of anything more than like a couple miles,

and I'm not even sure if I've seen that. They just want a broad range. And a lot of these are like limited to like a kilometer or so. This is very much like this kind of like over the shoulder, point an antenna at it, knock it down. It's a squad defense thing. It's like very or spot defense, point defense maybe you could call it. And it's for the small commercial stuff. to get something big, you could jam it and you could spoof it.

We have evidence of, I think there was for a while in Afghanistan, Predator drones were just broadcasting data over an unencrypted channel. And we found out that it was being watched by the Taliban. And I think we, and we know that at some point, I forget what years exactly, but Iran managed to spoof a, I think it was the RQ-170 drone down to the sky. and that's basically a flying stealthy wedge with cameras and other sensors in it. And they managed to spoof it. I mean, that was years ago. So there are ways that same technique can be done. The devices that we see as like the, I know if you were to print your modern Armyman drone, anti-dron unit guy,

those are going to be focused on quadcopters. So that was the, that was 2011, the RQ17 incident or RQ 170 incident. And you bring up an interesting point. if we can kind of transition into that. We usually think of drones as being synonymous with American military power. That's not always the case now. Who else is using them and how? I know we've got the kind of the off-the-shelf drones that we're seeing in groups like ISIS,

but who's got the actual big flying through the sky drones? So if we're looking for the actual big flying through the sky drones, the major drone powers, if you will, are the United States, Russia, China, Iran, India, Israel, and I think the UK and I think France kind of share the same models or roughly the same models in there. But of those, Israel has the longest history with drones, or the longest, especially with drones for what we would call against a right. drones in a regular warfare kind of thing. The United States had a big drone program in like Vietnam, and then we sort of abandoned it, and then we brought it back.

But Israel's had a pretty continuous program since the 80s, and they export, they export a lot. They have some populations, and I guess we would euphemistically say security challenges or more active captive security crises and opportunities is a gross word. But they use them a lot, and they market them a lot.

And they have ways to keep iterating their drone design. And they have big ones. They have like the long endurance kind. God, the euphemism, not the euphemism, the acronym for this, there's like a medium altitude long endurance is a family of drone, which is male. As if the drones weren't already so clearly that in appearance. But there's, so they do it. India has a small market. Iran makes some medium-sized ones.

We've seen Iranian ones. used in the conflicts currently in the Middle East, the including ones modified to be like explosive. They carry explosives and then they're one-way drones. And so that's somewhere between like a rudimentary guided missile and a drone or the term that is the like drone adjacent category of loitering munition. And that's another thing that Israel has a lot of, which is a drone that has a sensor on it, can fly around for a while. When it picks up a certain signal, it can figure out where that signal is coming from. and then it arms its warhead and then crashes into it blowing up. If it doesn't see that, then it can land and take off and do another mission out there.

But that would be the harpy or the harop. So there's a world beyond what the United States does with drones. It sounds like there's not really the same barrier to entry as with more sophisticated aircraft, though. I mean, it sounds pretty democratic. Iran doesn't make its own fighter jets, for example. Making a drone is in many ways a lot easier. Partly because there's like on the very small end, there's a huge commercial market. Four or four drones you can adapt and then you can kind of scale up.

And besides quadcopters, but there's also like lots of fixed wing drones that can do scouting things or be converted into explodey tools if you want. But one of the really like interesting things found as the United States and Iraqi security forces were pushing back through Iraq, reclaiming stuff. from ISIS is that they found ISIS had a fully like, ISIS hadn't built a war industry and they have the kind of things we expect, right? Like they had just munitions, they had like mortars. They were making their own ammunition. They were making some of their own weapons. But they also had multiple drone factories or maybe drone workshops is the better term.

But they were building quadcopters and fixed, when they weren't building quadcopter. Well, they had quadcopters, but they were building fixed twin drones. They could use with like a very simple propeller and very simple controls. you can see them. I'm like taped on antennas on like styrofoam or plywood bodies. And it looks really like a very like sloppy balsa plane scaled up and sort of mad mad scientist into existence. But they had that. They built their own drones during the war and they used them at least in the Battle of Mosul. I don't know if they use them elsewhere. But yeah, so there's a low barrier entry for like people for non-state actors and it's low enough for other nations without the same kind of legacy.

aviation companies or massive industry to still put something in the air that flies. The Pentagon likes to warn, well, they don't like to, but they will warn, or maybe part of their selling pitch, warn that the era of American air superiority or air supremacy is trending towards an end, the fact that we had decades reliably where American forces could go fight a war in a country and never worry about attack from the sky is over. The skill of that attack, the ability of the drones to cause harm or to reveal new danger, that's different. It's not comparable. No one's flying B2s or B1s or A10s or anything at us.

But there's something. The air superiority isn't the same as it used to be. That actually brings me to a point that I was thinking about. One thing that I'd read, and I think it was actually in a book called Ghost Fleet by Peter Singer and August Cole, They were talking, which is like very near future technology. And one thing they talked about was having jets that were piloted, having their own fleet of drones that are controlled from the air. Is that a real thing?

It's closer to being a real thing than you might expect. I'm going to just make sure I can get the name of these right. But I wrote last summer about there's a company called Kratos. that was working with the Air Force Retroats Lab, and they have two designs for drones. Kratos presently makes, like, target drones. The longest tradition of drone in military use is something that flies a pretty simple path,

and then the Air Force shoots it down, so they learn how to shoot things down in the air. But they make a drone called the Mako, which is a modified adaptation of the BQM-167 aerial target, and it's designed to essentially, you could command several of them from the from the cockpit of a fighter pilot or even and this is I think the more interesting case

you could have someone like in the back of a C-130 directing these and they can do things like they can fly it do maneuvers at up to 12 Gs was a big selling point right if you don't have a pilot on board you can turn in ways that would break a human spine and they can carry weapons and that's one that's like in the works like the Air Force Research Lab is presently partnered with them, and they're figuring out how to get

these things so that when traditionally, right, we think if we deploy a fighter squadron, and it's a squadron of F-16s, or it's a squadron of A-10s, or it's a squadron of F-35s. And in this case, it might be that you would deploy a squadron, a mixed squadron, of escort drones and a handful of fighters. and those escort drones will do a lot of the actual, like, fighting. They'll be commanded by the pilot. And I wouldn't be surprised if we see this within five years in some form. They also had a bigger one called the Valkyrie,

which can go faster, both can carry payloads internally. So that's a literally in the works as we speak kind of thing and might shift just how fighters work generally. Now, the scenario where that matters is less clear because the United States hasn't fought an Air Force that is anywhere close to it in capability. In decade, you could make the case that the Korean War was like the last time there was a real Air Force parody. And the Vietnam War was the last time that the United States was up against an Air Force that posed a significant and durable, meaningful threat. But that who knows what wars the future hold. So it might be relevant.

they might find a way to use it for the more standard kind of counterinsurgency air missions we have now. But it's a thing, right? Like it's a thing. And the last point I want to make on this is it's in part. Autonomy is what makes all of that work. The autonomy of the drones themselves flying, of them flying in formation, of them flying information without crashing into the human piloted plane that's directing them, the autonomy of the plane that the human is inside.

and piloting, maintaining its present altitude and everything while the pilot takes a few minutes or even just like 30 seconds to click a button on the tablet and direct the drones elsewhere. There's a whole lot of software doing the coordination and sensors, feeding the inputs needed for that coordination that make any of this possible. That leads me to my next question, which is, do you use? ever see this stuff operating autonomously at a level where humans aren't

pulling the trigger if they are if these drones are you know have weapons on them do you think that will ever allow them to make judgment calls about killing so I think so there's a couple ways to approach this and I don't think

we're ever going to see someone say this is the first robot we have decided to deploy that we are authorizing to make kill decisions. We're going to back into it, if anything, but we're likely going to back into it. Other countries are likely going to do the same.

And it's going to happen incrementally. One of the ways to think about it, I think, and this comes from Paul Scher's Army of Nunn book, which is another, like it came out last month. It's a really interesting look at autonomy in the future of war, is that a landmine is a machine that has full, it's unbound in time for when it makes a kill decision. It's very fixed in place, right?

A landmine will only kill a person in one spot. But once it's in place and until it's demined, the landmine will make a kill decision based on the input. And usually the input's like a person presses a button and that pressure is enough to make it all go cabloy. But that's an autonomous machine we have out there. And we have to look at it, I think, a lot in like dimension of time. He also cites the long-range anti-ship missile, which is a Lockheed project that was initially built as like it makes its own decisions.

And then as people like, wait, we don't want a missile making its own decisions. They've scaled back how to market it. But what it does is it's designed to be fired at chips far away. And there are some selected chips or some models of ships that it will be able to detect with onboard sensors and figure out. And so what happens is if it's launched and the first target it was supposed to hit is gone for some reason. It can't make it or it was already hit by something else. Then it will select a different target from a set of criteria inside. And there's a lot of opacity in this, right?

This is the kind of thing that if you're concerned about how machines are making kill decisions, this is one in the works and having some transparency of how is pretty important. but that's one that's likely going to make it through and be deployed. And there's a human involved in that process. The human initially picks the targets and the human shoots the missile. And then what happens in between the human firing the missile and the missile hitting a target is where the autonomy factors in. And in this case, the human press the trigger and then the weapon makes the call. The other example, or the other example, the other term we'd like to.

think of, right? And this is more relevant to drones, right? Say we have an autonomous Reaper-like drone flying over a target. It's using the cameras on board. It's processed images. It matches with some range of probability. A person of interest or a person in the United States has decided is a lawful target of war. And then it goes to execute the person. what it could do is it could be we have right now human in the loop kind of thing where and this is something that everyone's pretty clear on it before making the kill decision but after identifying the target and like tracking the target depending on one system generally the public seems to if to the degree that they are

comfortable with this this is the kind of comfort they have is they want to make sure that it checks back in that there's someone sitting at a desk someone trained someone with like a heart and blood in their veins who presses the button and says yep that's a target and I agree, or says, no, I don't actually think so, and calls it off. And that's the likelier way we will see autonomy, is we will see autonomy in target selection before we see it in target execution, and then it'll check with the human once it's acquired the target, and the human will say, yeah, sure. What will happen after that is we will see what's called human on the loop, where we have,

take that same scenario of the Reaper-like drone, it finds a target, the target matches the characteristics. It's been programmed. It meets all the internal coded criteria. And then it lets the human know that it's going to make the decision. And the human can look at it and say, wait, don't do that. But if the human doesn't say no, then the machine will go ahead and fire. And that's human on the loop, where the machine is monitored by a person rather than like the machine's process has to stop while waiting for a human. and we'll likely get there. We'll likely get there accidentally or inadvertently or incrementally,

but there's degrees of autonomy that are already present, and there are degrees that seem like logical next steps and how weapons are developed before... We won't just like great skip from machine gun to Terminator. There's going to be a lot of range in there, and it's going to be like that frog-boiling thing of, well, this was slightly more autonomous than the previous thing, and then slightly more than that,

and then slightly more than that. But to get to Terminator, we have to get to the ground first. We do have to get to the ground. Well, we could, I mean, yeah, the metaphor for autonomous flying machines is, I think, the 2005 cinematic experience of stealth, which no one likes and is a horrible example. So there's a metaphor there too, but, like, we do have to get to the ground, and there are ground machines we could be talking about. Well, let's talk about them a little bit.

What's coming to the ground and what's coming to the water? So here's an interesting one where the United States isn't necessarily first to field. We've known for a while that Russia is developing its own autonomous or its own unmanned ground vehicles, ground combat vehicles. There's a bunch of acronyms, and there's no convenient term like ground drones. So robot tanks, we could say, and Russia is developing several of them. And they announced last week that the Uron 9, which looks basically like, very small tank. It's got treads. It's got the little like armored body with the same like sloped hull. And it has a turret and it has a 30 millimeter cannon on the

turret which tankophiles will be very clear is not a big enough gun to count as a tank. But for literally everyone else, it's a tank. And they've been developing that. They've tested it. They have like videos of it like driving through like on hills and through snow and running through a wall that's on fire and things. But they announced that they had been testing it in Syria. And they announced it last week because it was going to be one of the vehicles in the big annual military parade. We haven't seen it in the open source in Syria.

It's possible they have been testing it and we just haven't caught it yet or maybe they've been testing it in a way where it's really hard to get open source footage of it. And we do know that Russia had the Iran 6, which is a demining vehicle, essentially a robot that you send to run over explosives or to flail at explosive. So the worst thing that happens is you have a broken robot instead of a mangled body. But the nine is the one we're going to be want to watch, because in other policy statements, Russia has said that they're hesitant to get on board with any of the laws about autonomous weapons because they don't genuinely know how they're going to work.

And what they would rather do is have field tests and then work from the law backwards, which for the perspective of someone who's maybe concerned about robots with guns, is insane. But that's one, and we've seen it. We can expect to see it in Syria now. And there will be more grand robots to follow. The United States actually tried to deploy a, well, and did deploy a, I think it was like Mars or something.

It was in the first half, the pre-surge half of the Iraq War, where we had a essentially a small-tracked platform like a bomb squad robot, but we just put a machine gun on top of it, maybe like a saw or something. And it, there were reports of it. accidentally like aiming at soldiers and that was enough and it was a whole scandal and that it was buried we have abandoned the attempt for now that we can expect to probably see further developments in something in the future and that's for the ground and the other big place where we can see it where we can see autonomous systems is underwater the biggest one of these the united states

has fielded a few different variations of of autonomous water vehicles there's a couple submarine types. I think Explorer and Ranger are the two big ones. And those just spend a lot of time underwater and occasionally like surface to send signals back and then keep going underwater. It turns out if you have a submarine, it's a lot easier to make it work for months if you never have to worry about keeping people on the inside alive. And so it's a great job for an autonomous machine with the massive caveat that it's really hard to communicate underwater. The signal's infrastructure that makes drones work and that makes ground robots work just doesn't go that far at all beneath water.

And I think I haven't read Ghosts to eat myself yet, but I think that's a pretty big part of it is that is how ships operate outside of this control. And there's a surface vehicle, too, that we should really be keeping an eye on called the Sea Hunter, which was started as a DARPA project and then handed over to the Office of Naval Research. And now, I think the Navy is just using it. And it's 130 feet long. It has, I think, room for like six people on top.

And it has a little like tower and like it has a couple life preservers on the side. But it's fully, it's fully uninhabited, unmanned, uncrewed. It's an autonomous machine that is designed to track submarines, to follow them, to persistently follow them and just let us know where they are. In the future, we might see armed variants, but even if we don't, it will have still been, like, to the sense that there is any, like, vague notion of fiscal responsibility and defense planning, sending a robot to follow a submarine for a couple of weeks is way cheaper than sending a, like, destroyer with a couple hundred people on board to do the same. And those are out there, and those are, like,

the U.S. versions I know off the top of my head, there's like smaller torpedo-shaped robots that do shorter scouting missions and then get scooped up by ships. We keep, I think, China captured one in those islands that they are building out into little forts. And we can expect, we can expect China, we can expect Russia, we have them. And I think, I don't know beyond that who's building underwater robots and water robots, but they are also part of this whole robotic picture. So really we can expect to see more and more of these things and, I mean, and we can't be entirely sure what comes next, right?

Right. Like, it's as maddening as it can be sometimes to, like, report on how I'm defending things in the U.S. is pretty good. I mean, a lot of respects about having, like, some public announcement of here is the machine we built and here's how it's going to be used. To the extent that I don't know what I don't know, sure. But like Russia, we have to go off entirely what their officials say, and usually after the fact, or when they have a very narrow thing, we don't know anything about the autonomy of the Iran 9. We don't know if it's like targeting on its own.

We don't know if it's, what censors it has. We don't know if it's going to check in with a human before it makes a firing decision. We do know that the range it operates is pretty limited. I think one unit is limited to three kilometers from the person controlling it. And I think a small group of them have a little bit more range they can work with, which is, I think, six kilometers. But that's still close. And those facts come from Scott Bendet, who does a lot of research on this over at CNA. But we don't know, like, beyond that it exists and that they claim it has some autonomous capability, what it does.

And we won't likely know until it's like captured in the software's analyzed or it's like publicly announced about it, which means we're sort of gradually sliding into a future where there are specifically like autonomous machines on the battlefield. And more broadly than that, it's not just autonomous machines on the battlefield, but it's autonomy in lots and lots and lots of other machines on the battlefield. field. One category two is that I think both Israel and South Korea have turrets where the gun is hooked up to sensors that identify movement. There's probably more filtering going on. I at least hope there's more filtering going on. But that's like a little bit of autonomy. It's in a fixed place, but that's still a machine looking for and tracking targets and making firing decisions and that those targets can often be human. And that's and that's just the the increasing automization of war is the trend. There's a lot, computers can process some decisions way faster than humans.

They can execute without panic or fear or moral qualms. The case, the pure military case of this thing needs to be destroyed or these people need to survive in the situation. how do I make my, what tool do I need that will best serve that need? There's a compelling case for autonomy. There's a lot of problems. There's a lot of limitations. There's a lot of dangers to say nothing of the humanitarian objections. Or just concerns of what happens when it can't be controlled.

What happens when the data link is lost? What happens when someone fools the sensors? The more autonomy there is, the greater the danger. of things getting hacked is of just to spoof or to confuse it. And the greater danger of like, what if we code in the laws of war as best we understand them and we mess up doing that or the people and robots we end up fighting against in the future don't have those same qualms built in?

I would be remiss to talk about this whole field if there wasn't also like active movement to preemptively regulate, under international law and the coal nations to not even use them. The stop killer robots is the big campaign in this space and there are others. And so to some extent, it's a policy debate and to some extent there's a, like there's a

the technology iteration and evolution. And to some extent, it's a really, like, it's just a big political question about the shape of war in the future. And while it's being talked up top level, we're going to see it filter out on lower levels all the time, I think. Thank you for kind of an amazing summary of where we are when we're going. And it's great to know that the killer robots really are coming for us. So we appreciate that.

It's good to know that a lifetime of science fiction novels and television shows and movies has prepared me for this inevitability. The great reveal behind the mask is that the killer robots were always on the way, but it was people all along. Yeah. No. Right. No.

He scored at the sky. Skynet decisions are coded by people. Kelsey Atherton, thank you so much for coming onto War College and walking us through this. Absolute pleasure. Thank you guys. Thanks for listening to this week's show. If you enjoyed it, leave us a review on iTunes or wherever you get your podcasts. Just like Jack Scratch 128 did.

Very informative and not what some might expect. Five stars. In-depth news and history on timely subjects. Thank you, Jack Scratch 128. We really appreciate the thoughts. You can subscribe to War College wherever you get your podcasts. And you can reach us at facebook.com slash war college podcast. Thanks very much for listening.