Drones and AI on the Battlefields of Ukraine with Kate Bondar and Sam Bendett
46:50 Share

Welcome back. I'm Max Bergman, director of the Stuart Center and Europe-Russia-Eurasia program at CSIS. And I'm Maria Snegovaya, senior fellow for Russia and Eurasia. And you're listening to Russian Roulette, a podcast discussing all things Russia and Eurasia from the Center for Strategic International Studies.

Hello everyone and welcome back to Russian Roulette. I'm your host, Maraisa Negovaev, recording solo today while Max is traveling. Today we have two fantastic guests on the show, our CSS colleagues Kate Ponder and Sam Bandit. Kate is a fellow with the Wadwani AI Center here at CSS, where she studies the integration of artificial intelligence into battlefield systems with a particular focus on the war in Ukraine and the Ukrainian defense industry.

Sam is a Senior Associate and Resident Fellow with our very own Europe-Russia Eurasia Program CSAS, in addition to his work as an advisor at the Center for Naval Analysis. Sam's research focuses on Russian defense and technology developments, uncrewed robotic and autonomous military systems, artificial intelligence, and Russian military capabilities. Sam and Kit, thank you for joining us. Let's jump right in. Thank you. Hi, thank you. All right. Kit.

I'd like to start with you. Earlier this month, you released a very interesting new report on css.org titled Ukraine's Future Vision and Current Capabilities for Waging AI-Enabled Autonomous Warfare. I encourage all of our listeners to go check it out when you get the chance. The link to the full piece is in our episode's show note. It opens with a notable quote by a

On the battlefield, I did not see a single Ukrainian soldier, only drones. I saw them, I mean Ukrainian soldiers, only when I surrendered. And there are lots and lots of them. Guys, do not come. It's a drone war.

Also, a more recent quote from Z blogger Ivan Otrakhovsky: "Jones now calls 70% of enemy wounds per Russian medics without air support an effective EW . Our guys die. The Kremlin plays those fine, but the front says otherwise." So, Kate, let's start from the broader perspective. Could you please comment on

this rule of drones in this war. Is it a real game changer? And how did the three years of war affect the evolution of the drones warfare? Yeah. Hi, Maria. Sure. I'm happy to speak about this because I really think that drones changed the current war and warfare drastically. And the number that you just mentioned, like 75% of Russian soldiers injured by drones and wounded by drones,

Yeah, I actually went and checked that article, which was published in Russian Medical Journal, and it's like 75%. And we can see different numbers, which are like all of them are striking. From Ukrainian side, when I was listening to the interview of the head of Ukrainian Unmanned Systems Forces, a new branch of Ukrainian Armed Forces, he mentioned that Ukrainian army

conducts 80% of its strikes on the battlefield with drones. And, you know, I think this number is really striking and it actually reflects the changing nature of modern warfare. So yeah, and it's actually a common impression of anyone who's been on the battlefield because there is no direct combat between infantry units anymore.

So we won't see like typical classical war as we think of it when we think about the war, because now it's all, let's say, remote and forces keep in the distance. And this is kind of remotely controlled battlefield. Of course, there is a huge innovation race and there is a constant competition between two sides on improvements in innovation, both in hardware and in software. And

If we're talking about software improvements, they happen even faster than hardware because usually it takes a couple of weeks for each site to reverse engineer or improve hardware part and the platform. But it takes even a couple of hours to upgrade and improve software. When we're getting to the software part, it brings even bigger changes to the battlefield rather than hardware developments.

Because AI-enabled systems, for example, AI-enabled software improves the performance of a usual drone like three or twofold. So for example, if we take a usual FPV drone remotely controlled by an operator, the success rate in striking by this drone is 10% to 15% because it depends a lot from skill set of the operator and a lot of stress factors on the front line, et cetera. But if you take exactly the same drone

and you enhance it and improve it with autonomous navigation, AI-enabled navigation. If you improve it with automated target recognition, also AI-enabled. And if you add to this solid AI-enabled last-mile solution, it almost becomes, you know, autonomous weapon system. And the success rate in engagement of these systems jumps to like 70% to 80%. So from 10% to 20% to 70% and 80%,

and 80. That's a drastic improvement as well. And if we take a look a bit on a broader kind of context of these changes, I would say these are more interesting trends that we can see over there and implications of these changes are very different and pretty unexpected and let's say unpredictable. So for example, what I'm saying is that the drone warfare changes not only tactics and

operational level, but it changes the war strategy, how wars are conducted, how operations are conducted. It switches the current war from manned to unmanned. And we have heard about fully unmanned operations and missions conducted actually from both sides. So we're trying, like Ukrainian forces are trying to save people's lives and remove humans from direct combat.

and replacing them with robots. And that's kind of sci-fi that we were thinking about a couple of years ago, but now it's actually true. It's not autonomous systems, they're still remotely controlled, but it's robotic battlefield, actually. And for example, another interesting implication of this change in warfare is that the psychology and psychological effects of, for example, drone operators, they also change a lot.

So Ukrainian forces have shared with me a very interesting trend and a scary trend is that, for example, drone operators, they are not directly placed on the battlefield. They don't have this direct sense of killing. And for them, it's like a video game. And you don't have the sense of killing someone directly, but you have this sense that you want to chase the target and you want to destroy the target, whatever it takes.

And it becomes like a game for you. And now the commanders, you know, unit commanders,

commanders, brigade commanders, they sometimes complain that they cannot stop their operators from chasing the targets. And they don't understand why this is happening. And they actually feel that they need to strike any target just not to waste the drone. And this is something that is scary. And there is no regulation for that. There is no law for that. There is no responsibility, basically. And this is a completely new...

effect that we're just starting to witness and we haven't even explored it. So what I'm just trying to say, drones have changed the tactical level and operational level, but it also has way broader implications. Some of them are kind of not explored yet, not very well understood, not analyzed. And I think in future, we'll see more and more of this effect.

effects coming into picture. Fascinating, and I guess really so many avenues for future research. Sam, can I please bring you in to comment on this topic also? How much are the drones a game changer? Are we really witnessing the war of the future, basically? We are, and I want to add to Katarina's excellent comments that this war is a mirror image. So

If one side gains an advantage or fields a technological solution, it is often copied by the other side or countermeasures quickly developed to that solution, which prompts more iteration on the solution in general, which prompts more counter solutions, so on and so forth. And so both sides are referring to it as the constant never-ending race of the sword versus shield. I also agree, and this has been stated numerous times by both sides, that

The technology in this war changes quite rapidly. And so anywhere between two to three weeks and up to three months is about the average span of a certain technology. There's an example I often cite. There's a Russian military company in Tula, which manufactures an FPV drone. And what they do is they put...

a gingerbread cookie inside the box with the drone when they send it to the front. And if by the time the box with the drone gets to the soldier, the cookie's stale, it means that it took more than three weeks for this technology to reach the front, in which case it's probably obsolete and needs to be updated or needs to be discarded altogether.

Another interesting aspect that Katerina has touched upon is the psychological aspect. The gamification of this war has proceeded at a very rapid pace, and now even Russians are sounding alarm that their own pilots are treating this war as a game, and especially when it comes to younger people, younger soldiers, younger pilots who are digital natives and who grew up with digital technologies.

Killing the adversary is not an ethical or moral question. It just becomes a thing they do, just like if they were playing a game. And so there are going to be probably long-term psychological and ethical consequences of this for both sides. But again, it's something that we're not really fully understanding. Another interesting aspect of this war, something that has been

on full display since 2022 and has been discussed on numerous occasions, is the rapid and unprecedented involvement of civil society on both sides in providing and developing technologies, especially tactical drones, counter drone systems, electronic warfare systems, unmanned ground vehicles, unmanned surface vessels, and everything and anything in between. And as the civil society step up, the line between a true combatant

and the civilian is starting to get blurred. And as the war continues, this line is blurring even more, especially when some of the volunteers are setting up shop near the frontline or at the frontline to directly assist their respective units and commanders. And finally, I want to say that this front is still very, very large. And so the developments that we're seeing

the developments that Katerina described in her excellent report and that we have covered on a number of occasions here, they're uneven across the front because there's no standards as of yet. And so both

militaries, Ukrainian military and the Russian military, are trying to enforce some sort of standard so that the technology that is manufactured and delivered to the front functions the same way everywhere. Right now, there's a lot of commentary, especially in the Russian language telegram channels, where people are saying how oftentimes the stuff that gets to them doesn't quite work, the stuff that gets to them from the volunteers or technical startups

or private sector efforts. And so there's a drive right now by both militaries to kind of take more of a control over how these technologies are fielded, how they're tested, how they're evaluated, and how they get to the front, which brings me to my last point, and that is the constant tension between this rapidly moving, rapidly evolving high-tech sector that is mostly housed amongst the volunteers and civil society that can quickly make necessary changes to whatever is needed

and whatever feedback they get from the soldiers, and the standard military-industrial complex, which is still investing heavily in legacy systems and cannot innovate and iterate quite as fast. And so this tension is on full display, and some efforts like Ukraine's Brave One platform are trying to mitigate that and trying to make sure that there's a better cooperation and connection between the military and

And the volunteers, Russians, are trying to copy that as well, but they're moving at a slower pace because it's a much more hierarchical and larger military force. Thank you, Sam. Fascinating. And it reminds me actually of the early days of Internet when traditional media was trying to kind of handle everything that was happening online. And now we have the same situation between the two parts of the military.

Kate, can we then maybe deep dive into your fantastic report? Could you please sum up for us the key takeaways that we haven't yet touched upon? Like, what do you think is the crucial thing for our listeners to know about the development of AI and the use in Ukraine? Yeah, so I have already partially started, you know, talking about the takeaways in my intro comments. And I think that Ukrainians have actually developed

a very good approach to innovation development. It wasn't like a formal strategy that they gathered and developed this genius approach. It appeared organically as a response to this constantly evolving need and existential need, basically. So what I'm saying is that Ukrainian government, first of all, decided to take a role of an enabler rather than a developer. So Ukrainian government didn't...

invest that much of resources into developing state-owned industry and trying to compete in drones warfare and making the state-owned industry to produce these cheap drones. And yes, these initiatives came from civil society, as Sam has mentioned. And that was a really smart move to actually outsource commercial technology from private sector and basically create a streamlined

pathways for this commercial technology to get into the military. So Ukrainian government has really improved and simplified testing and evaluation procedures and then procurement procedures. So previously, this whole process of a new technology getting into Ukrainian arsenal took like five years or six years to go through testing and evaluation, then contracting and actually producing the

product and delivering it to the frontline. And this happened during the war, which started in 2014. And now this process takes from two weeks to up to a month. So this is a very fast route and way for commercial technology to get into the hands of end users on the frontline. Second part is that Ukrainian manufacturers realize that it's really hard for them to compete in developing

the whole system, so basically hardware and software. And that's why Ukrainian company is focused on developing separate parts, which they call modules. So if someone is producing a frame and a drone as a hardware, someone has focused on producing separately autonomous navigation systems, basically a software, which they deliver as a small little module, which can be integrated basically into any drone.

Other companies focus on ATR, automatic target recognition, but it's AI-enabled, which means they take a computer vision model and they train it on the data set that they collect, and it's able to recognize objects and classify them and make targeting more efficient. Others focus on last-mile solutions, so the systems become resistant to localized electronic warfare and all types of jamming, spoofing, etc. So Ukrainian...

Industry applied this modular approach, and basically it's easier to address current challenges and specific challenges on separate parts of the frontline. Because as Sam mentioned, the frontline is very long. It's really huge. And the situation is very different. And that's why demands and requirements for weapon systems are very different.

So, there are parts where there is stalemate, there is nothing, no fighting is happening, and you just need ISR drones just to monitor the situation and keep it under control. But on the other parts where there is active hostilities, you really need bombers, you need kamikazes, you need ISR drones. And when you get this flexible system, easily customizable system, basically a frame, a

put all elements that you need, you can easily adjust your drone to the specific tasks and needs that you have specifically in this mission or that part of the frontline. Another part, which again, Sam has already touched on, is this standardization of practices and experiences. And Ukraine has also made, I think, a very smart move in creating this unmanned system forces. And

And many people here are asking, like, what is it? Like, is it a special unit or is it like a separate, it is a separate branch of armed forces, but do they have their own forces? Do they have their own units, brigades? They join others or like, how does this work? Basically, they don't. They have a very high skilled drone operators who join different units on the frontline and they guide them. They learn from their experience. So basically, they're trafficking

traveling from unit to unit and they collect this experience, they analyze it and they standardize it. So they're actually fulfilling this function of experience exchange and that's how they develop like best practices and they upgrade and upskill other units, you know, like what others do. So it's a very efficient system to address the challenges, to exchange experiences and

And they help to distribute drones more efficiently just to satisfy the need which exists on that part of the frontline. So basically, there's three big parts of government standing aside and just being an enabler and helping this commercial technology to get into the frontline, communicating requirements to the industry through these different platforms, hackathons, challenges, etc.,

This is what is really helpful. Experience, exchange, and creating standards on best practices of unmanned systems usage. Of course, there's flexibility and modular approach, which actually makes systems easily adjustable. And it saves time, it saves resources, and it makes the systems more affordable to a bigger amount of warfighters on the front line. I'll probably stop there because I can talk for quite a long time about this. Well, we'll definitely get back to that, Katie.

Kate, is that right, by the way, that the training and learning about using this type of systems, operated unmanned systems equipped with autonomous features, may take as long as 30 minutes, meaning that basically our listeners could have learned this during the duration of this episode, right? Yes. Well, this 30 minutes fact and statement, that applies to already skilled drone operators who already know how to operate the drone. They just need to adjust to using these systems and

Basically, yes, this is just like 30 minutes to show when you give control of the system to the software and then you take it back. So how this works and how to manage this switch, all the rest is. But in general, yes, it really shortens the time of preparation of a drone operator because previously it took approximately three months of high-quality training. Now with AI-enabled software and autonomous software,

navigation, for example, using this type of drones, the training can take up to a week because what AI does, it takes control over the drone functions like the drone flight, obstacle avoidance, adjusting to wind gusts. So previously, the operator had to do this all manually. And now it's like using an iPhone and pressing one button and the system reacts by itself. So this software makes

systems more available to a bigger amount of soldiers on the front line. Fascinating analysis. Sam, I'd love to bring you in perhaps with some comments on some of the points that Kate made. And also, if you made flag from your perspective, especially given your familiarity with the Russian side of the story, what is different? Like, how do you see maybe the Russian army and the Ukrainian army differently approach the usage of drones? Is there anything like pronounced enough there that's worth mentioning?

Well, I think my previous comment that this is a mirror image still stands. In December of last year, Russians announced their own creation of the unmanned systems forces, and that force should be stood up sometime by October, November of this year. And it is likely that Russians are going to see what Ukrainians have done, and they're probably going to copy and paste the best

approaches, like with the Ukrainian force. The unmanned systems forces in Russia is probably going to be centered around some of the more capable drone units in operation, like the Rubicon Center,

and similar initiatives. And they will try to incorporate lessons learned and spread the lessons learned and make this learning more horizontal. As far as actual learning by pilots, I think last year Russians were claiming that they could limit the FPV pilot training to no more than three weeks of intensive seven-day awareness.

week courses, at least eight hours a day. And this involved a number of related initiatives like technical knowledge about the drone, drone repair, cooperation with electronic warfare and signals intelligence, including tactical medicine, how to do first aid, and so on and so forth. What's interesting about Russian training is that they are now

spreading it across a number of government-funded, regionally-funded, military-funded, and volunteer-funded initiatives. So they are racing to deliver thousands and thousands of drone operators to the front because drone operators are the number one target in this war. Both sides are trying to locate the adversarial drone launch sites, crews that launch UAVs or bring supplies to the

drone units and target them before they do anything else. Again, Russians are looking at Ukraine's best approaches, but Russian military was busy trying to understand the impact of AI for years before the invasion in 2022 and even before the

2014-2015 invasion of Crimea and eastern Ukraine. And so the Russian military has been heavily invested in artificial intelligence research and development. They're trying to codify and understand what that means for their own forces. And again, to distill this down to what

what Russians are saying about it right now publicly, artificial intelligence is going to aid in data analysis and decision making. And just like in Katerina's report, Russian initiative also aims at fielding a number of AI-enabled systems in war and

And of course, there's a big difference between writing about it and actually fielding it. Right now, on the Russian side, the artificial intelligence is fielded in a limited capacity, such as target log, target acquisition, image recognition technology as a way to bypass electronic warfare. But Russians are also busy researching the impact of war and vast collection of data

on how their military systems are going to function. And so Russians are also trying to gather as much information and data as possible from an absolutely staggering number of data points present in combat, whether it's satellite-based, drone-based, soldier-based, or social media-based. And so Russians are going to use this vast quantity of data to train their systems

to develop data sets for artificial intelligence, research and development for aerial systems, ground systems and maritime systems. But in many ways, Russians again are following what Ukraine has done. Russians have not fielded a

successful number of unmanned surface vehicles. So far, they've been talking about it, they've been researching it, they've been testing it for years, but it was the Ukrainian military that was the first mover. When it comes to unmanned ground vehicles, we're not witnessing Russians field any sophisticated, exquisite military concept that they were working with before 2022. For example, the majority of unmanned ground vehicles in this war are cheap, tactical, attributable systems.

similar to an FPV or quadcopter concept. Something that's easily built, easily lost, and easily replaced.

But again, Russian military has been investing in certain types of capabilities for quite a long time. And we have to keep in mind that just because they haven't been talking about it or just because they have not made something public doesn't mean that it doesn't exist. And so we have to watch what Russian military is discussing to kind of discern the patterns in their research, development, testing and evaluation cycle.

Thank you, Sam. And may I clarify, in our previous podcast where we discussed this issue of drones, you've been mentioning, if I remember correctly, that basically the drones have made this frontline transparent. And does it mean that it's also easier to learn from each other in terms of the way drones are applied just because the frontline is now so transparent? Well, absolutely. There are still different tactics used by both sides.

There are still new capabilities which are being developed. Drones now fly much farther than in 2022 and 2023. So this gray zone from the line of contact now can extend up to 20 kilometers. Further in the rear, 30 kilometers, 40 kilometers, which prompts both sides to change how they conduct logistics, supply, how they conduct

conduct military operations and any other related function. So there's still a lot to learn. And of course, Russians are trying to utilize new tactics, especially in their assaults. One of the newer tactics recorded in 2024 was

small assault units bringing FPV drones with them all the way up to essentially a closing position, essentially the short distance before they close in with the Ukrainian positions. And then they essentially launch these FPVs, which are then piloted by pilots in the rear, which gives Ukrainian soldiers or any defender very little time to defend against. So this is a relatively new tactic, which has been enabled by, first of all,

advances in communication, advances in signal strength, and sheer number of FPV drones which are available to the Russian forces, which are also available, of course, to the Ukrainian side as well. We're talking millions of drones that were flown in 2024, something that we probably could not conceive of when the war has started, and certainly something we probably never thought about before 2022.

Again, I have to add that when we talked about military UAVs or UAVs in a war, we were mostly dealing with military forces fielding military-grade UAVs in smaller numbers because these were multipurpose drones for intelligence, surveillance, combat, and related missions. Now we're essentially talking about civilians building these drones in the hundreds of thousands each month.

and the knowledge in building and fielding these UAVs has spread far and wide. And so this is something that we have to consider as well. What happens years from now when both societies have so many people with so much knowledge about combat? How are they going to be applied? How are they going to be integrated into the civilian workforce or into the society in general?

So many new ideas about protesting the government, for example. That's scary. That's actually scary. Yes, it is scary indeed. So since we're already on the topic of how the drones are affecting the current situation on the battlefield, Keith, maybe I can bring you back to talk a little bit about that.

Is it even possible for the frontline now to use, given that drones basically are doing all the work and every side is quite aware of what the other party is doing? Also, maybe to what extent drones are a substitute for labor shortage and manpower shortages in the frontline, particularly problematic for Ukraine, but for Russia as well, to some extent? And lastly, the role of the reconstitution. Do we even have to bother reconstituting other lost

armed systems while maybe we just build up more drones and they substitute for everything. What's your take?

Yeah. So as I said, Ukrainian forces are conducting their strikes with unmanned systems and 80% of strikes are conducted with unmanned systems. 70% of Russian equipment is destroyed with drones. So the numbers speak for themselves. But the current stage of technology development doesn't allow to substitute people completely. And some systems which are more complicated, they

They require actually two people to operate them, so like one drone and two people. And this is where we're expecting the technology development to get into the stage when we can deploy swarms of drones where we have the opposite balance. We have one drone operator or two people operating a swarm of drones of hundreds, for example.

But again, it requires a bit different level of AI development. So drones can actually collaborate in between each other. It's not the air shows that we see today where it's pre-programmed, it's algorithmic flight. It's not the collaboration and coordination in between drones. So we're not there yet. So we don't have full autonomy on the battlefield. We don't have it deployed from any...

any side. And that's why this battlefield is still heavily manned, although it's not the direct participation of warfighters in the combat. But yes, we're moving into that direction. And also, there's a constant debate about can we replace legacy systems, big conventional systems, expansive systems with new, treatable, cheap systems. And well, I can speak for Ukrainian

armed forces because I have a lot of interviews with them, and they still consider that artillery cannot be replaced with drones yet. Ukrainians develop different tactical approaches and new tactics, how to combine unmanned systems with artillery and make these missions as efficient as possible in terms of target engagement and target destruction and in terms of cost. Because when we talk about this engagement efficiency and

and success rate. And we calculate, for example, one target destruction by artillery shell, which costs us like $5,000. And cheap FPV drones, which might be already not that cheap if we enable them with all the software, et cetera. So they might cost like $2,000. And if you need two drones to strike one target, so it approximately equals and it's approximately the same. So in every mission, there should be a balance of

tactical reasons and justification and probably cost efficiency and safety reasons for the personnel, et cetera. So for now, I wouldn't say that unmanned systems completely replace conventional systems, but they definitely add and they make capabilities more and capacities more diverse and

If you just come up with some creative tactics, how to deploy them and combine them in a mission, that would actually increase the mission success and cost efficiency. And these are two main factors anywhere. And I would also like to kind of point out and stress, we're talking about unmanned systems, about drones, you know, in all domains, but there is also a very important part where

where AI and new technology can actually improve the battlefield situation is situational awareness. And this is where AI is super efficient, super effective, and it adds really a lot to situational awareness and as a result, mission success. So for example, current battlefield is filled with sensors, all types of sensors, because all these unmanned systems are equipped with cameras and

and they transmit a lot of information to the center, to military headquarters, where this information should be analyzed and add to common situational awareness system and add to common operating picture. So AI actually helps to analyze that information. A lot of information and intelligence is coming from text information. These are group chats of intelligence and reconnaissance units. These are reports of civilians.

These are chatbots, intercepted communications of the other side. And we also need to accelerate this process of analysis and make it more efficient to approach situational awareness to real time or as close to real time as possible. And last but not least is acoustic system. So Ukrainian battlefield is equipped with microphones placed as a grid along the front line. And Ukrainian command center receives

This analysis of acoustic signals coming from different parts of the front line, which helps to project what kind of threats are coming, are these rockets, are these drones? And all this information combined, analyzed with the help of AI, contributes to this situational awareness and approaches it to real time. And knowing what's happening on the front line or behind the front line is a vital need and

it really adds to proper planning, mission planning, and strategic planning. So we shouldn't forget about this, and especially when we're talking about unmanned battlefield, because we want these machines to work almost independently from people and from drone operators, and they need to get information from somewhere. So the ultimate goal and kind of big dream

is to combine these two big aspects, situational awareness combined with collaborations with machines. So basically, based on this real-time information, machines can conduct their missions on the front line. And this is where the Ukrainian military is moving towards. And this is where AI technology is indispensable, essential, and critical. Although you can also imagine how this can potentially backfire given the scale of the change.

that we are not really able to control. Sam, back to you, along the same lines, would you please comment on the current situation on the frontline being affected by the drones? Can we even see the frontline move now? Is it even possible? And what are the preconditions for it? Also, how do drones substitute for labor shortages and the reconstitution rates of other systems?

Well, we have an example of Ukrainian attack against Russia's Kursk region as sort of a lessons learned and a combined sort of final outcome of Ukrainians learning about Russian capabilities. In essence, Ukrainians were able to

counter Russian ISR and communications along that sections of the border. They were able to go after Russian ISR drones. They were able to jam Russian communication frequencies. They were able to take advantage of some of the confusion that resulted amongst the Russian forces and the fact that the Russian forces that were deployed on the border probably had

less experienced. That may have been a very unique experience and a unique moment in this war, but Ukrainians took full advantage of that, and they poured into that breach, and they were able to hold on to Russia's territory for a very long time. So this is a lesson for other militaries to go after the adversarial communications, electronic warfare, and ISR capabilities, ground-based or air

air-based ISR capabilities. If that is possible, then the front can move and a limited assault can be launched. The question then becomes what happens afterwards when the adversary learns very quickly and tries to counter your initial advantage, which is what we've seen in Kursk. And one of Russia's

advantages against the Ukrainian forces was the fielding of a fiber optic drone on a large scale. So once Russians started fielding those UAVs against Ukrainian forces, it became very difficult to interdict this type of a UAV. Russians didn't have these drones in large numbers. Once they did, it became possible for the Russians to interdict and affect

Ukrainian supply, communications, and logistics lines going into the Kursk region. And so there are lessons that we can still take from what is going on at the front right now. Both sides are obviously trying to come up with ways to overcome the adversarial defenses. I will add

on the fiber optic control drone, we talked about civil society in this conversation. So Russian fiber optic UAV did not originate at a large defense industrial corporation. It actually originated from a volunteer-facilitated effort, which essentially built itself as the

technical accelerator. And so this largely volunteer effort was able to field a groundbreaking technology at scale, which was then copied, of course, by the Ukrainian forces and which was then scaled up by the Russian forces themselves. So now both sides are rushing a lot of FPV-controlled UAVs into combat. It is possible, again, to overcome your adversarial

capabilities, but you have to be very specific how you want to do it, where you want to do it, and this probably is likely to happen after a specific weakness is identified in a defender's position, capability, equipment, or how the defenders in that specific location are using certain technologies so that a rapid breakthrough can be affected and then build upon.

I will add on the manpower. Obviously, drones are extremely important, but drones do not take the territory, right? The

The territory still must be held by infantry. And so drones are extremely important in treating and weakening the adversarial capabilities. But then the infantry still has to go in and secure that specific location that has been won over. And so drones can do that. We don't have mechanized androids quite yet, although we do have some commercial developments right now that have recently made news where androids

android, humanoid, robotic systems look like people, move like people, and can potentially perform tasks like people. Again, that's still in development. In wars like Ukraine, you still need infantry, you still need regular troops and regular soldiers to go and take territory, hold on to towns, and do everything to maintain hold over a specific part of the front.

And that's not going to change anytime soon. And so both sides are now getting ready for a confrontation and for the phase of the war where regular ground forces have to communicate with and collaborate with and function well with the robotic systems, ground and aerial, that will augment regular infantry's operations.

Fascinating. Thank you, Sam. Everything that you've been both describing so far really sounds like an episode of Black Mirror or maybe some sort of sci-fi movie. Last question, because I want to be cognizant of our participants' time. What about the next steps on

on the battlefield. What is there in terms of the innovation to come? Is there something you're specifically looking for? Sam, I understand you mentioned the ability to break through the defenses of the enemy. Maybe anything in particular that you think might be a real game changer in both tactical and strategic terms? Kate?

Yeah, so I see the future in software improvements still, because even fiber optics drones, this is like, I take it as an intermediate measure just to combat shortfalls of software and AI-enabled systems for now. But I still think that we will see a lot of improvements there, and AI systems will become cheaper as well and more affordable.

So they, like Ukraine last year purchased 10,000 drones out of 2 million drones with AI-enabled features. But this year, you know, Ukrainian military plans to buy half of drones with AI-enabled software. So we will see broader deployment of AI-enabled systems. We will see more direct communication machine to machine with situational awareness, with ground and

aerial and sea systems, cross-domain collaboration, and cross-domain operations. And yeah, the future with humanoid robots, or we don't know what kind of robots is there going to be, but definitely this is what is coming because a human, to be honest with you, is a really expensive asset and not a reliable asset in the war because it's really easy to kill. It has emotions, it has stress, fatigue, etc. So

The warfare will move to autonomous systems in all domains, and there will be more and more machine-to-machine collaboration. Thank you, Kate. And Sam, anything to add to that?

I agree that there will be more machine-to-machine cooperation, integration of autonomous and unmanned systems in combat. I think it's worth pointing out that Russians were the ones writing about this sort of coming age of automation in

in war. Russians were describing in their military periodicals, and a lot of very smart people from their military industrial complex, as well as from the military academies, were pointing out that the rapid spread of commercial technologies is going to impact how militaries will integrate automation and unmanned systems and

how soldiers will be pushed back from the war, and then Russian invasion happens. And we essentially saw the return in many ways to World War I, World War II style tactics enabled by UAVs, enabled by advanced systems, both at the tactical, operational, and strategic levels.

I think the way out of a positional deadlock is a combination of short-range, mid-range, and long-range strikes by a number of unmanned systems followed by an advance by the first wave, if you will, of

certain types of augmented military technologies or autonomous systems, which are then followed by regular infantry and mechanized infantry. That may seem like science fiction or it sounds well on paper. How that's actually going to play out in actual combat is impossible to predict. Again, we did not know three years ago we would be here now. We did not know three years ago we would be talking about drones in this way.

We did not know three years ago we would be talking about the volumes of different types of autonomous systems, and especially UAVs, that we're discussing right now. And three years ago, no one really considered an FPV-type drone as an important military tactical asset. They existed in the commercial space, but in a very limited format.

Same goes for a commercial quadcopter. They existed already. They were there for wedding photography or agriculture or infrastructure monitoring, but no one really considered them at scale as a military asset in the numbers that we're discussing today. So I think

A lot of developments in future combat can take place over a very short amount of time and can mature very rapidly. We can either build on what we have today and utilize what we have today, or something else will come along and will completely upend our understanding of how we fight, what we fight with, and the numbers of military assets that we are fighting with.

Thank you, Sam. And I think I feel like it's a good place to stop being cognizant of our even failures of imagination, basically, despite everything that's going on. Fascinating topic. And I really feel like we only scratched the surface today. But I wanted to thank our fantastic participants today, Kate and Sam, thank you very much for taking time to speak with us this morning. And it's really a fascinating topic. Thank you. Thank you. Yes. And of course, you're welcome. It's

It's been a great conversation and to all our listeners, thank you as always for tuning in and be sure to subscribe to our show and please give us a five-star rating if you haven't already. Lastly, please do not forget to check out our sister podcast, The Europhile, wherever you get your podcasts. See you next time. You've been listening to Russian Roulette. We hope you enjoyed this episode and tune in again soon.

Russian Roulette releases new episodes every two weeks on Thursdays and is available wherever you get your podcasts. So please subscribe and share our episodes online. And be sure to check out all the latest analysis by the Europe, Russia and Eurasia program at csis.org.