Systems that support reconnaissance, predict enemy movements, or even autonomously conduct combat – seemed like a futuristic dream barely a few decades ago. Today, artificial intelligence is joining the army all over the world.

The military training ground in Nowa Dęba, May 2024. The FlyEye drone takes off. It flies over enemy-controlled territory, records an image, and streams it to the command post in real-time, where it is instantly analyzed. Artificial intelligence (AI) algorithms help to recognize, identify, and determine the force and position of foreign troops, model their possible movements, and develop operation options for the command. This is how the modern BMS Legion works. It was created by the specialists at the Command of the Cyberspace Defense Forces Component (DKWOC). Currently, work is underway to introduce it to various combat platforms. “The system is one of the foundations, based on which the Future Task Force concept has been developed,” explains MajGen Karol Molenda, Commander of Cyberspace Defense Forces Component. The FEX-24 Exercise was a great opportunity to test it.

Dirty, Boring, Dangerous

The idea of the Future Task Force (FTF) project was born four years ago in the General Staff of the Polish Armed Forces. It was simple – to form sub troops for testing innovative solutions in equipment and armament for soldiers, but also tactics on the battlefield. The experimental unit was formed as part of the 21st Podhale Rifle Brigade. “It was formed by the soldiers of the 5th Battalion of Podhale Riflemen,” explains Col Marcin Zalewski, Deputy Commander of the 21st Podhale Rifle Brigade. Initially, the FTF officers and NCOs took part in computer-aided command and staff exercises. Afterwards, the troops operated on the training field.

Computer scientist John McCarthy.

During two FEX exercises, soldiers tested, among others, futuristic drones. One of them was the PIAP HUNTeR – a nearly 4-ton unmanned ground vehicle (UGV) for observation and fire support of light infantry, or a universal land carrier (ULC) designed to transport a load of several hundred kilograms. Artificial intelligence was equally important. “For now, we apply it mainly in the image recognition processes. We verify to what extent it can relieve specialists and support the command. With the AI, data analysis is incomparably faster, which is of considerable significance on the battlefield, where seconds decide about success or failure,” emphasizes Col Marcin Zalewski. However, he soon says: “AI is a powerful tool, but... only a tool. During exercises, AI doesn’t make any decisions for us. These are always at the competence of a human.”

Generally speaking, systems tested during the FEX are still a matter of the future. However, the Polish army already uses devices that benefit from AI, such as the loitering ammunition, which after firing autonomously follows the target and precisely hits it with the help of AI. Such systems will probably arrive at an avalanche pace. As experts unanimously emphasize, AI is the future. “The Cold War brought us the development of technologies that are today defined as first-wave. The invention and spread of nuclear weapons, ballistic missiles capable of carrying warheads over long distances, and satellite systems were a breakthrough. The 1990s have brought second-wave technologies: precision weapons, modern command and control systems, as well as stealth solutions, such as bombers difficult to detect by radars. We had the opportunity to feel their power during the First Gulf War. Artificial intelligence is sometimes classified as third-wave technology. Its appearance has opened a completely new chapter for us,” emphasizes Col Rafał Kasprzyk, PhD in Eng., Deputy Dean of the Faculty of Cybernetics at the Military University of Technology (WAT) in Warsaw.

Until now, modern technologies have been developed mainly with the funding of individual countries, largely with the army and defense in mind. From there, they were transferred to the civil market. Recently, however, there has been a symptomatic change. Private businesses are now also working on the development of AI. “The army must keep up with the rapidly changing world, implement and develop often existing solutions. This requires a wide range of professionals,” says Col Rafał Kasprzyk, PhD in Eng. The game is certainly worth the candle. When asked in what aspects related to defense AI can be useful, experts instantly reply: practically in all of them. Recognition is one of many. “Solutions with AI algorithms can not only quickly process data from various sources but can also help with geolocation. AI can detect the location of a given object, based only on an image, without any additional metadata,” explains Col Rafał Kasprzyk, PhD in Eng. AI can also be applied in logistics. “Calculation algorithms will help to precisely define the needs of given troops or tactical units, and optimize delivery times and transit routes. In addition, autonomous vehicles controlled by AI will make the loading and unloading of supplies simpler,” explains Radosław Rudek, PhD in Eng. from the Military University of Land Forces (AWL) in Wrocław. Col Rafał Kasprzyk, PhD in Eng. adds: “In the past, Stanisław Lem said that artificial intelligence would relieve a human being in all tasks that are dirty, boring and dangerous. It is easy to imagine AI-controlled machines that transport heavy loads, and also clear mines off given areas, and participate in masking strategic objects or building crossings. In short, they are used in an area related to the broadly understood protection and survival of troops.”

PIAP HUNTeR, an unmanned land vehicle weighing almost 4 tons, can be used for observation and fire support of light infantry.

There is one more issue – cyber security. “Artificial intelligence will allow you to more effectively secure communication networks. It can quickly detect anomalies in the data flow that indicate a cyberattack,” explains Radosław Rudek, PhD in Eng. Last but not least, combat systems: AI can improve the control of existing weapons, and also... replace a human on the battlefield. Until a few decades ago, considerations about self-learning systems, intelligent drones, or robots firing at enemy positions would sound like science fiction. Today, it is slowly becoming a reality.

China for X-Ray

In mid-December 2021, the US Indo-Pacific Command released information about an innovative tool it allegedly uses. With the help of AI, the tool can predict with great precision the reaction of Chinese authorities to various activities taken by the United States in the Indo-Pacific region. The system is to operate based on data collected over the years and related to, among others, China’s attitude to the sale of weapons from the USA to Southeast Asian countries, American military activity in this part of the world, but also congressmen’ visits to sensitive places, such as Taiwan. Kathleen Hicks, then Deputy Head of the US Department of Defense, commented that they were aware that analysis of a wide range of index rates and how they are interleaving was necessary to understand interactions between threats. Since its launch in 2020, the system, as its users assured, has assessed the activities that have affected the relations between the USA and China, and indicated potential strategic frictions. With this in mind, the US decision-makers knew their limits with Beijing and were able to avoid decisions that would lead to an escalation of tension between the world’s hegemons.

The problem is that the information about this tool is scarce. To this day, it is not known under what name it operates. Of course, we can accept this information is top secret, however, it is impossible to rule out that the 2021 statement was a sort of a trial, a test balloon, which was supposed to... check China’s reaction. Either way, this example shows one of the options for AI use in the army. With AI, it is possible to predict scenarios based on historical data. The secret American system is supposed to create models useful on a strategic level. At the same time, in the US Army (and not only there) algorithms for designing enemy actions at the operational and even tactical level are very intensively developed. That’s not all.

The Americans used AI, for example, during their search for Osama bin Laden. The system collected and analyzed data on the movement of people in the buildings under observation. It was important who and when is entering those buildings, under what circumstances is leaving them, in which rooms, and at what times the light is turned on. Eventually, the services gained information about the purpose of a given object. AI is also more often used for the so-called predictive maintenance or predictive maintenance of military equipment. In short, it is about engaging artificial intelligence to monitor tanks, aircraft, and ships, to help accurately determine when their components will need to be replaced, for instance, due to material consumption. Example? The PANDA (Predictive Analytics and Decision Assistant) system was developed for US military aviation. It uses AI to analyze enormous data packages. What’s more, with machine learning algorithms, the system can modify its knowledge, detect new dependencies, in short: increase the effectiveness of operations. PANDA, by monitoring the condition of aircraft, not only detects anomalies but also predicts potential problems with the equipment before failure even occurs.

Technological Twins

Meanwhile, European armies are already working on similar solutions. In the spring of 2024, at the request of the French Ministry of Defense, scientists began implementing the NumCo project. It assumes the creation of something like digital twins for VBCI infantry fighting vehicles. While operating, they receive data collected from sensors installed in a small, test batch of machines. As a result, AI creates the model of how a given vehicle functions, and, in a way, predicts its future. This way, the exploitation process is under control, as AI determines the life expectancy of components or helps to minimize the risk of failure. If the project goes as expected, France will gain a tool that can be used to support other types of vehicles in their armed forces, as well as other French machines produced for foreign contractors.

There are, of course, more similar examples. World powers are working hard to develop AI-related initiatives and research programs. Currently, the USA is leading the way in this race, but they are very much chased by the Chinese. A decade ago, Beijing announced a plan to become a world leader in AI by 2030. A similar document was prepared by NATO. The AI development strategy was made public in 2021. A year ago, the Alliance introduced updates to it, which consider the dashing technological progress and the complicated international situation. That’s not all. The European Union and individual member states conduct their research on artificial intelligence. The West has long concluded that excessive dependence on foreign solutions and technologies can lead to disaster. It is enough to mention the COVID-19 pandemic and problems with semiconductor supplies. In addition, there is the awareness that in the political dimension the world has entered a completely new era. China already openly questions the Western value system and arduously strives to redefine the global order. Russia takes a similar position, only that it decided to take shortcuts, invading Ukraine and intensifying hybrid activities aimed at NATO. The Kremlin, like its opponents, has recently prepared its own AI development strategy.

It seems that in the coming years, the collective security system will be increasingly dependent on digital solutions, about which a few decades ago only futurists discussed.

What Does a Robot See?

Artificial intelligence is a double-edged sword. On the one hand, it enhances opportunities, on the other hand, threats. “AI enables the automation and autonomy of complex attacks, which can be used by enemy forces to disrupt defense systems, manipulate information, and take control of critical infrastructure,” emphasizes MajGen Karol Molenda. “For example, deepfake technologies allow you to create realistic but false images and recordings, which can introduce information chaos, misinformation in peacetime, crisis or conflict, weakening morale and trust in the information delivered by intelligence and the media,” he adds. That’s not all. The tools that are supposed to serve us may also become a problem. To put it simply, it is a matter of how much freedom on the battlefield we give to the products we create. Experts are talking about three concepts here. The first one, named ‘Human in the Loop,’ assumes that a human remains in a decision-making loop. In other words, a human being fully controls AI-powered systems. AI only develops and suggests some solutions, but the final decision – on whether to use a given weapon and how to use it – is always taken by a human.

The second concept is called ‘Human on the Loop.’ This, in turn, assumes that a human gives machines much freedom. However, he controls the operation of the systems at all times, and if something goes against his intentions, he can interfere with the whole process. Finally, concept number three – ‘Human out of the Loop.’ According to this idea, a human is completely excluded from the decision-making process, fully relying on AI. According to experts, this is an insanely risky concept. If only because AI is capable of self-learning, and the decision-making processes generated by it remain unclear to humans. In other words, a machine can behave in a way that a human would never act.

“Suppose the task is to control a given territory. Still, what does it mean? How to do this? Who will guarantee that a machine, allowed to decide on its own, will not for instance conclude that the execution of an order requires moves that are completely unacceptable to a human?” asks Col Rafał Kasprzyk, PhD in Eng. Kaja Kowalczewska, PhD, a lawyer at the University of Wrocław (UW), who deals with the impact of new technologies on international criminal law and the modern security system, is of a similar opinion. “When developing AI and striving to optimize certain solutions, we tend to ignore how complicated the battlefield environment is. Today, without having to worry about legal or ethical issues, we could only use it if armed conflicts were taking place in the desert or on the moon, and not in densely populated areas. Which is pure fiction,” the expert believes.

Sometimes even the use of AI in the role of an advisor raises doubts. Such was the case with the tools that Israel used in the war against Hamas. The Lavender system, analyzing, among others, biometric data from the inhabitants of the Gaza Strip, typed potential military targets. The representatives of the Israeli army assured that this was only a tool, the data of which is verified by humans. “The problem is that the number of attacks carried out after Lavender’s recommendations was very high. This leads to the conclusion that the analysis could not be too in-depth, because the officers responsible for the system would simply lack time for it,” admits Kaja Kowalczewska, PhD. In her opinion, similar cases should initiate wide discussion and lead to new international legal regulations. However, none of the global powers are interested in imposing a rigid framework on itself when it comes to AI use. Echoes of similar discussions also reach Poland.

Although Poland does not own such advanced technological capabilities as the world’s largest powers and does not even have such funds as they do, it has still decided to focus on AI development. Poland develops its scientific background in such a way that it can meet the latest challenges related to the implementation of AI. “Students of the Faculty of Cybernetics at the Military University of Technology (WAT) learn to build command support systems, control systems for combat means, and support of all information and decision-making processes. We are working on the use of AI in managing different types of drones, or in CyberOps and InfoOps operations. At the same time, we cooperate with various centers, including of course with the Cyber Defense Forces Component Command (DKWOC),” says Col Rafał Kasprzyk, PhD in Eng. AWL is working on the use of AI to support ground troops. “In our area of interest, there are mainly unmanned aerial vehicles,” admits Radosław Rudek, PhD in Eng.

In August 2024, the Polish Ministry of National Defense announced a comprehensive strategy for the implementation of artificial intelligence. The program covers the period until 2039. At the same time, Władysław Kosiniak-Kamysz, Minister of National Defense, and Krzysztof Gawkowski, Minister of Digital Affairs, signed an agreement on the establishment of the AI Fund. “We will build a large institute in Poland to deal with artificial intelligence,” Gawkowski emphasized. Poland plans to allocate a billion zlotys for this purpose. The Artificial Intelligence Implementation Center (CISI), which is being built at DKWOC, will also be launched soon. As MajGen Karol Molenda assures, it will be a milestone when it comes to the implementation of new technologies in the Polish Armed Forces. “Artificial intelligence will play a key role in the way military operations are conducted in the future,” the Major General believes. “The new institution will enable the rapid integration of AI in various domains, such as autonomous combat systems, cybersecurity, intelligence analytics, or logistics,” he says. As he adds, both army specialists and civilian experts will work for CISI. “We plan to implement pilot programs on AI, as well as to modify educational programs for military studies. This will enable some outstanding graduates of WAT or civil universities to serve in the Center,” concludes MajGen Karol Molenda.

In short: new times are coming. On the one hand, AI makes it possible to strengthen security systems. On the other hand, it is quite tricky, and its traps can be relatively easy to fall into. To effectively avoid them, we should not even for a moment forget that AI has been created by a human being. As such, it is a human being that should control it at all times.