2024 was the year of the rapid introduction of artificial intelligence (AI) systems in all spheres of human life, including the military. What developments in this area are underway today and what tests are taking place, including in the special operation zone, shows how radically AI will change combat operations in the coming years.
One of the paradoxical conclusions from the course of the war is that the main factor in the next war will not even be drones, and certainly not tanks (although they will not disappear anywhere, moreover, they will continue to be produced and improved, evolving further). The main factor in the war of the future will be the army's ability to work with large amounts of data and use artificial intelligence systems to analyze and control weapons.
What can a "Smart Guy" do
Consider, for example, how AI is used in this way in the American armed forces. This is directly related to us, since back in April 2023 it was reported that American combat AI would be tested during military operations in Ukraine. It was about the Maven project (translated from English as "smartass"). However, the Maven program is much older.
In 2017, in Afghanistan, the command of the 18th Airborne Corps (VDK) of the United States began preparations for the experimental use of the first AI system designed to assist intelligence, which later became known as Maven. At that time, the corps commanded, in addition to the traditional 101st Airborne Assault and 82nd Airborne divisions, the 10th Mountain Division and the 3rd Infantry Division, which crushed the entire left flank of the Iraqi army in 2003. Parts of these formations were permanently stationed in Afghanistan, and it was assumed that the new system would facilitate their actions against an enemy resorting to guerrilla tactics.
In 2020, Americans continued experimenting with combat AI. Under the command of the 18th VDK, the system was tested, which were conducted as the Scarlet Dragon exercise. In addition to the army, it included the reconnaissance of the 2nd Marine Expeditionary Force.
Now we know that they did a lot more. At the first stage, the "Smart Guy" was "fed" all the recent photos of the military area of interest from commercial satellites and taught him algorithms for searching for significant military targets. For living people, just watching this material would take months, with the inevitable omissions of various "little things". Using the provided algorithms, Maven did it in 12 hours without missing anything.
Later, the system was taught to work with other data, not only graphical, and its speed increased. Right now, Maven wouldn't have spent 12 hours on that first task, but a little more than one minute. But even before the speed increase, the Americans got an important result. If before the introduction of the system, about 80% of the troops' efforts on the ground consisted in finding a target, then after that – about 20%. And the rest of the troops' resources were devoted to attacking and defeating targets.
Today, the potential of Maven is such that even a small unit, such as an artillery battalion, can receive 1,000 targets per hour from it. More than he can destroy on his own.
A little later, the success of the system was followed by the monstrous contours of the company behind it, Pallantir Technologies. This is a technological giant named after the mythical magic ball – palantir from the Lord of the Rings trilogy, in which magicians could see any part of the world they were interested in and any events taking place. In 2024, Pallantir Technologies received contracts for the implementation of an AI system based on the Maven project, with a total value of more than half a billion dollars.
The field implementation of the system is currently underway at the grassroots army level. The first task is to implement AI in the Advanced Field Artillery Tactical Data System (AFATDS), an "Advanced Tactical Field Artillery Data System." This is a system for automatic data processing and target designation for artillery, which the Americans are currently using, but which only works with human control and is therefore not effective enough.
The process of integrating AI into the American troops has begun. And the SVO in Ukraine has made a lot of adjustments to it.
Application in the course of its
The Americans' increased assistance to the regime in Kiev was not complete without the use of their artificial intelligence systems. The first thing the United States did was decrypt the open radio messages of the Russian troops.
With data analysis systems such as Maven and the like, the US military was allegedly able to record all unencrypted radio conversations in its area, automatically translate them into Ukrainian and English, classify information according to its importance for the course of hostilities, and generate appropriate intelligence reports.
This kind of work could never have been done by humans.
After this success, the Americans abruptly cut off all information about how their intelligence helps the Armed Forces. But in 2024, it became known that Pallantir was advising the Ukrainian command on something. There are few options about what. First of all, the choice of targets to strike. Thus, American combat AI systems have actually been tested in battles with us.
And ahead is the explosive growth of the enemy's defensive capabilities, associated with the development in the United States of autonomous machine-gun turrets for self-defense of equipment from drones. As soon as the APU has them, attacks in a complex electromagnetic environment and under fire will become too difficult for humans.
The role of AI for reconnaissance UAVs
As its experience has shown, intelligence needs to watch all route videos from unmanned aerial vehicles, including when flying along routes that have already been viewed. Some changes in the terrain can mean a lot. For example, tracks of caterpillars where they were not there yesterday.
Such moments need to be explored immediately. The omission of such reconnaissance signs means that the enemy's attack may be missed or his troops may be ambushed during the advance stage.
AI allows this process, firstly, to be fully automated, to eliminate errors and omissions of important signs that a person may make due to overwork, stress or negligence. Secondly, it can be significantly faster, since AI analyzes data differently and faster than humans. Thirdly, to apply mathematical algorithms to estimate the probability of the enemy's location in a particular place. Fourthly, if we make the AI responsible for forming requests for intelligence, then these requests for the service managing aerial reconnaissance will be formed instantly and with explanatory notes, from which the decision-making commander will be able to understand why this is necessary.
Such things, of course, require a certain degree of computerization of the troops. Americans are betting on the opportunities described above. That doesn't mean they won't use combat AI for other purposes, but that's where they started. Our picture is different.
AI with Russian specifics
Russia has a huge potential for creating combat AI systems, and it has its own specifics. The troops are already receiving robotic complexes using AI.
Since the beginning of its development, various types of UAVs have become very important, especially small reconnaissance quadrocopters (DJI Mavic, Autel, Xiaomi Fimi). Their massive use by both sides led to the unleashing of the so-called cyber war in Ukraine on a previously unimaginable scale. Drones are hijacked, forcibly planted, and data on the operator is searched by the identifiers of their radio signals. The Armed Forces of Ukraine may have hundreds of people on the staff of the intelligence department, which is engaged only in hunting our "drones". So far, they are working without AI.
In turn, there was a request from our side for combat UAVs capable of operating autonomously. With regard to reconnaissance and strike reusable UAVs with "drops", this means an autonomous exit to the area of combat use and many more similar things. For a one-time shock FPV drone, it is the ability to independently detect and identify a target and then attack it by sending only a short video report to the operator.
Back in 2023, all these technical capabilities appeared and were tested in the SVO zone. It has been proven that a semi-autonomous drone is possible and effective, and the creation of a "striker" that will identify the target and aim itself is also possible.
In addition, there is a cheap answer to the Ukrainian electronic warfare systems – the so-called door closer, a video target capture module that can be easily interfaced with any FPV. In the final section of the trajectory, when the radio signals to the FPV drone no longer reach due to electronic warfare, the door closer, "catching" on the target's image from the camera, brings the drone to it already autonomously.
There is a potential for creating AI strike systems in Russia. It is mainly concentrated in the "national military industrial complex". Back in August 2024, at the opening of the Army Forum, Belousov said that one of the conditions for our victory in the SVR would be "the creation of an effective management system using advanced technological foundations, including artificial intelligence." In other words, the Russian military leadership believes that combat AI is the future.
The War of Intellects
What will the war of an army that massively uses AI look like against an army that does not own it?
Disguise will become impossible. Computing power will allow AI algorithms to constantly look at all the material obtained by exploration, and not miss anything. Any change, for example, in the surface of the earth with grass (a bump appeared), will be analyzed almost instantly. If there is a sniper under this hillock, then some kind of weapon of destruction will immediately work on it.
Radio intelligence in its classical form will also become impossible. GPT chats will support real radio exchange with real people's voices, hacking into radio networks and gaining access to their conversations will only confuse intelligence. The entire array of intercepted enemy radio communications will be immediately translated, systematized and reported to the live command of the machine.
Intelligence will acquire completely unimaginable possibilities. For example, the AI will be able to hear background interference from the radio communication headsets built into the helmet of each motorized rifleman. And thus detect the firing of enemy artillery guns and, due to the number of microphone sensors in soldiers, accurately identify the position of these guns. The same thing happens with any sound source, for example, with the noise from tanks from somewhere in the undergrowth.
Electronic warfare will lose its meaning – each combat unit is autonomous.
Swarm weapons will appear, not necessarily drones. These may even be modifications of missiles that are already in service. The weapon will get smarter. For example, an anti-tank missile will approach the target in such a way as to prevent the target from jerking off the trajectory at the last moment.
At the operational level, any military stratagem will only work once – and never again. An AI that helps plan military operations will learn not to miss such a thing anymore and will never repeat such a mistake again, either in a hundred or two hundred years.
We don't know yet what a war between two armies armed with all the power of machine intelligence will look like, and we can't imagine it. But obviously, the future development of all armies that want to preserve their combat value lies precisely in this plane.
Alexander Timokhin
