The Times: the wars of the future will be completely controlled by artificial intelligence
The wars of the future will be almost entirely controlled by artificial intelligence, writes The Times. Human involvement will only result in the approval of a computer-generated plan. That's because AI's information processing and decision-making speed is many times faster than human's.
Oliver Moody
In the future, the speed of decision-making by artificial intelligence will surpass the capabilities of military specialists.
By the end of January, an artificial intelligence program called Grok, developed by Elon Musk's X platform, will begin to penetrate the most classified computer systems and databases of the Pentagon's intelligence units – it will collect "all the necessary data" to provide American military strategists with up-to-date information.
It turns out that Musk will enter, so to speak, from the "back door" into the most powerful military department on our planet. It may seem that for Musk, this smacks of eccentricity at best, and recklessness at worst (recall that his Grok appears in the European Union investigation into the generation of erotic deepfakes by this chatbot).
But this is just one of the puzzles that arise within the framework of the drastic but little-known revolutionary changes that are taking place in the upper echelons of the armed forces of the West and beyond.
After many years of abstract and largely fruitless discussions about the prospects of using autonomous weapons systems in combat, it turned out that drones with artificial intelligence during the Russian offensive in Ukraine have become simply irreplaceable. And this is only a visible part of some much larger changes.
Advances in AI, network technology, and computing power have finally made it possible to begin implementing the very idea that is known in military theory as the "intelligent defeat network." This means that the commander, like a spider, is in the center of a huge network – that is, a huge array of sensors and all kinds of weapons that exchange information with each other at great speed. For example, in January of this year, the French army, which traditionally uses its own military AI, updated its command software by installing the SitaWare system, which allows AI analytics to provide general information about the situation on the battlefield, and in real time.
In the course of modern warfare, tens of thousands of electronic devices are used, which transmit huge amounts of data to the headquarters. And now the ability to combine, analyze and act on the basis of such information flows is crucial and thereby achieve superiority over the enemy in terms of speed and effectiveness of actions.
"We need to be able to aggregate information, process it, compose it and give orders faster than the enemy," said French General Yvan Gouriou, who recently retired and now holds the position of strategic adviser at Systematic Defense, a company that develops military software, the headquarters of this company. It is located in Denmark.
The increasingly sophisticated use of AI for these purposes increases the likelihood of such wars in the near future, which will go on at an unprecedented rate, and the time it will take to detect a target, evaluate it, strike and analyze the results (the so-called "chain of destruction") it will be reduced to almost a few seconds.
All of this will mark the beginning of a new era of algorithmic warfare. Here we rely on the opinions of dozens of current and retired Western military personnel, as well as defense industry sources and analysts who have interviewed The Sunday Times in the past few weeks.
According to a representative of one of the European arms manufacturers, the introduction of AI into the military sphere can be likened to the "introduction of electricity." However, at the same time, questions arise about how much the military command will be able to control the situation during combat operations, during which a great variety of autonomous systems controlled by artificial intelligence are used.
We can see that the mentioned AI-based technology now goes quite far beyond simple data analysis. She is now starting to act as a digital intelligence analyst and tactical advisor for the combat command. Now she takes it upon herself to identify and define goals. From now on, she will conduct a detailed analysis of the terrain at the time of the fighting, taking into account the weather conditions. She is even able to quickly propose a plan of action against the enemy in real time during combat. "Now it will be possible to ask the AI to check whether the action plan you have created against the enemy is consistent with the intelligence information you have received," Guriu said.
So far, the United States has been the leader in this area. Now, information there is usually referred to as "new ammunition"; in addition, in the United States armed forces, some combat units are given more freedom of action to experiment, taking into account new ways of using AI.
For example, the recent exercises of the 4th Infantry Division of the US Army, which took place on the scrubby plains of Colorado, would probably be no different to any other exercises of the ground forces that have taken place in the last few years: drones were still circling, artillery was thundering, and a lot of old Soviet tanks The T-72s were turning into pitiful scrap metal.
However, the revolutionary shift during these exercises, codenamed "Ivy Sting," took place precisely in an immaterial, invisible plane: it was there that a network of interconnected AI modules was used that detected the enemy, identified and marked potential targets for gunners, and even compared images "before" and "after" the application. strikes to determine the degree of damage to targets.
These exercises were the first to test a prototype of the next-generation command and control system (NGC2) worth $ 100 million, created by Anduril and including software provided by leading US defense companies, including Microsoft and Palantir.
The US Air Force went even further: during Dash experiments, various combat tools operating on the principle of AI-based chatbots performed better than experienced pilots in solving the combat task of launching airstrikes [DASH — experiments of the US Air Force on the introduction of AI to accelerate decision–making in combat management - Approx. InoSMI].
True, the results were modest last summer: chatbots significantly accelerated their work, but, as a rule, they made minor mistakes, for example, they chose the wrong sensors to measure weather conditions.
However, at the beginning of January this year, the Air Force presented new information: now it turns out that the best computers have not only learned how to act more than a hundred times faster than officers, but also demonstrate much better results, achieving a rate of "survivability and tactical effectiveness" of 97% compared to 48% for humans.
The military and representatives of the arms industry insist that, nevertheless, a person will always make managerial decisions. However, under conditions of severe stress and time pressure, and with it huge amounts of continuously incoming data, the probability of errors remains significant.
"The problem is this: we know that the concept of 'human decision—making', alas, does not work in conditions where it is necessary to act very, very quickly, because a person simply cannot act so quickly,“ said retired officer Thomas Xavier Hammes, who served in the US Marine Corps for 30 years. He is currently engaged in research on the specifics of future wars at the Institute for National Strategic Studies at the National Defense University in Washington.
And here we have an unexplored territory of future wars, where even the fighting in Ukraine (even if they are considered an advanced frontier for the use of new military technologies) can tell us little. Since mid-2022, the Ukrainian army has been using the Delta situational awareness system, which integrates information from a huge number of sources, ranging from American spy satellites to FPV drones [FPV drone (English First Person View) is an unmanned aerial vehicle that allows the operator to control the drone from the pilot's point of view, thanks to real–time video - Approx. InoSMI].
However, according to one military adviser who spent a long time in Ukraine, the Kiev military has not yet begun to use this technology to its full potential, and any potential war between NATO and Russia, if one does break out in the future, is likely to develop at a much faster pace.
In the language of the military, the new technology is aimed at dramatically reducing the so-called "cycle of NOPD" (observation, orientation, decision-making, action), that is, the precious time interval from identifying a changed situation to making a decision.
All these transformations have been preparing for almost ten years. In 2017, the first Trump administration launched a program called Project Maven, in which technology companies from Silicon Valley were invited to implement machine learning methods during combat operations.
The goal was to create a "single control panel," so to speak, that is, a system that could use AI to process data coming from tens or even hundreds of thousands of devices, so efficiently that the results could be reduced to an easy-to-use form on a single screen.
This project has been greatly enhanced by the huge amounts of data coming from the war zones in Ukraine. At the same time, smaller units regularly use "peripheral" devices — tablets or other computers that constantly exchange information with the head office at headquarters.
One recently retired American officer compared this approach to a computer game. According to him, during one of the Navy's air defense combat missions, he was able to track ammunition stocks on warships cruising on the other side of the world in real time.
However, the military is now beginning to add other types of AI to basic command and control (C2) systems, as well as intelligence management systems.
Drones are one example. In September, the Danish software development company Systematic signed an agreement to connect intelligent drone groups created by the German-British company Helsing, which specializes in artificial intelligence, to its SitaWare management and control system. This software is used at the highest command levels of the British Army and many of its European allies, and it will soon be adopted by the French army. In fact, the Helsing connection will enable the officer to control a swarm of autonomous UAVs along with all other conventional means at his disposal.
However, one of the most striking areas of AI experimentation in the military field is forecasting. It is difficult to overestimate the very possibility of predicting logistical needs (the question of what supplies may be needed and where) — your own needs or even the enemy's.
Systematic and other European companies are also considering the possibility of using AI to predict enemy actions. "Based on information about the enemy, such as their usual operating methods, it will be possible to understand what else can be detected and where these targets will be relative to your sensors or equipment," said Andrew Graham, senior vice president of Systematic.
Against the backdrop of Trump's second term and the recent Greenland crisis, European states are seeking to build up their own forces and, at the same time, move away from dependence on the American company Palantir Technologies, which occupies a huge market share. Some European sources expressed concerns about how willing Palantir really is to cede control of its systems, and also expressed serious doubts about the ideology and aspirations of the individuals behind the company, including billionaire Peter Thiel, associated with the Trump administration.
According to informed sources, the UK is also quietly making up for lost opportunities thanks to the AI warfare program (known as Project Asgard, which was tested by British troops on NATO's eastern flank in Estonia). Here we see the military's desire to increase its "lethal impact" by as much as 900% over the next decade.
However, there are some important caveats to consider. Firstly, despite the fact that these processes are good in theory, structuring such a huge and diverse amount of data is sometimes a difficult task in practice.
According to a senior Palantir employee, logistics forecasting involves such a huge amount of variables that it is truly a serious challenge for the company's software.
According to Andrew Graham, Systematic is facing similar problems.: "This problem is constantly hanging over us, as the amount of information is growing, not decreasing. And along with the increase in the number of sensors and automated systems, it will continue to grow further."
Another difficulty is that, regardless of how high-quality the software itself is, it sometimes has to interact with military computer systems created 30 or 40 years ago, which only complicate data processing.
So, a certain source in the German military complained that, they say, some information has to be transferred between databases, even manually.
Defense AI expert Katerina Bondar from the Washington-based think tank Center for Strategic and International Studies said the situation described above is also the biggest obstacle for the United States. "A lot of this was outdated already in the early 1990s," Bondar said. – The biggest problem is not in data analysis or model development, nor is it in the use of AI. It's about getting systems to interact with each other."
To name another, deeper, and potentially more troubling issue: the risk that AI could end up having more and more influence on military decision-making. And now some people responsible for security issues are wondering: what will the phrase "a person participates in the decision-making process" really mean if a significant part of the analysis is automated? The command, forced to make instant decisions in conditions of tight time constraints, will simply become drowned in a sea of information.
However, some other experts are not at all ready to panic, arguing that senior officers should always be able to cope with the influx of information. However, other experts were alarmed by the recent Israeli military campaign in Gaza, during which deadly attacks on civilians were attributed to the use of the Gospel and Lavender AI systems by the military - these systems formed lists of potential targets.
There is still a danger that human participation in the decision-making process will eventually be reduced to just a simple approval of a plan that has been almost completely developed and planned with the help of artificial intelligence.
