MOSCOW, April 29, by David Narmania. «Whoever becomes a leader in the field of artificial intelligence will be the ruler of the world,» Vladimir Putin said six years ago. Since then, AI has come a long way, including in the military realm. About the application of these developments — in the material.
«Dead hand»
The authoritative National Defense magazine published an article where he spoke about the «living laboratory for AI». We are talking about the conflict in Ukraine.
One of the most developed areas where artificial intelligence is used is reconnaissance and target recognition. For example, as the publication notes, it is used to determine geolocation from photos from open sources. Or face recognition. The program used by the Armed Forces of Ukraine was created by the American company Clearview AI.
However, when it comes to AI, it is very difficult to say exactly where its «field» is. The computer in a broad sense has been used by the military for a very long time — automation, for example, helps to identify targets and direct missiles at them. But he is capable of more complex — and sinister — tasks.
Russian system » Perimeter», created back in the Soviet Union, was dubbed the «Dead Hand» in the West. This is a complex for automatic control of a massive retaliatory nuclear strike.
In a war of annihilation, as you know, control centers are the key target. The task is to decapitate the enemy troops and prevent them from reacting. That's how Perimeter works. Having found signs of a nuclear strike, the complex sends a request to the General Staff. If no answer is received, he turns to his «colleague» — the Kazbek strategic missile control system. It is she who, in a nuclear suitcase, is carried by an officer in naval uniform next to the president. If she does not respond, «Perimeter» itself makes a decision. But for this, its sensors must detect a nuclear bombardment. Only after the first strikes are ballistic missiles launched. In other words, revenge is guaranteed.
However, even such a complex system, according to experts, is not yet artificial intelligence.
As Vasily Burenok, Major General, Doctor of Technical Sciences, notes in an article for Military Thought, the automation of command and control processes in the military sense involves the use of computers to collect information, classify it and further make decisions by people. But in the case of artificial intelligence, the computer itself makes decisions under conditions of uncertainty, when not everything is known about the ongoing events. queue in order to dispel the notorious «fog of war». One example is the Pentagon's multi-million dollar Project Maven initiative.
The project should become the new «eyes» of the military. A person is not able to process a huge amount of video and photographs from American satellites and reconnaissance drones — that's why they attracted an assistant. «photo» data-crop-ratio=»0.666666666666667″ data-crop-width=»600″ data-crop-height=»400″ data-source-sid=»ap_photo» class=»lazyload» width=»1920″ height =»1280″ decoding=»async» />
It is with this project that the first high-profile scandals around ethics issues are associated. Employees of Google, which acted as one of the contractors of the Pentagon, filed a petition in April 2018 in protest against the company's participation in the creation of military technologies. The letter was signed by about four thousand people. When there was no response, several dozen of them quit.
Nevertheless, a few months later, the IT giant refused to renew the contract with the US Department of Defense. The contract expired in March 2019.
The cost of a contract for such a corporation is low: Forbes reported $15 million. However, Google's place was taken by competitors — Microsoft and Amazon. They earned a total of 50 million and want to continue to cooperate with the Pentagon.
Learn to fly on their own
They did not stop at target recognition. Already in 2020, the United States signed a contract to equip MQ-9 Reaper drones with artificial intelligence. General Atomics received $93.3 million for this.
Now the company is developing a technology that will allow UAVs to go offline — to fly independently, select tracking objects and process data. In the future, they want to teach artificial intelligence how to control a swarm of drones. Developments in this area are already underway.
Death from heaven
However, the control of individual UAVs is also being improved. Last week, the Russian company «Unmanned Vehicles» presented a prototype of a jet-powered FPV kamikaze drone of the K-5 aircraft type. It can recognize thermal signatures and visual patterns. In other words, the operator must bring it to the target area, and the K-5 will find it and destroy it.
At the same time, some samples are already superior to humans in controlling drones. According to the South China Morning Post, in March, the Chinese military simulated a battle between AI and human-controlled strike teams. =»photo» data-crop-ratio=»0.75″ data-crop-width=»600″ data-crop-height=»450″ data-source-sid=»» class=»lazyload» lazy=»1″/
And the computer did not leave a chance for a person — despite the fact that at first he seized the initiative using different tactics. But artificial intelligence quickly corrected the situation, being able to go behind the opponent. The operator could not break away from him, although he maneuvered for a minute and a half. This is an eternity for a battle, the outcome of which is sometimes decided in a split second.
Commenting on the experiment, the researchers express the opinion that relying on aircraft capable of hitting targets from afar is a dead end branch of the development of military aviation.
«Thanks to the technologies of stealth and electronic countermeasures, 25-40 percent of air battles in the future will be fought at close range,» experts emphasize. artificial intelligence is already fully armed in battles. And we are talking about full-fledged combat aircraft.
In 2020- The Defense Advanced Research Projects Agency (DARPA) of the US Department of Defense held the AlphaDogfight Trial competition. The participants — eight development teams — created a program capable of controlling a fighter. They tested the brainchild on the FlightGear simulator, which simulates the F-16.
In the end, their creations fought each other. The winner was an AI from the Heron System company, which then clashed with an experienced Air Force pilot. The confrontation ended with a score of 5:0 in favor of the program.
New Challenges
But it is one thing to control a drone, a fighter jet or even a swarm of drones, and another to fight. In the current conflict in Ukraine, UAVs already allow monitoring the situation, but people are still leading the process. In the future, it is proposed to transfer this function to artificial intelligence.
And it is already gradually being involved in command and staff exercises.
The ability of AI to lead units better than humans was confirmed a few years ago. In 2019, Elon Musk's Open AI program defeated the reigning world champions in the Dota 2 game, the OG team. And in open access matches, when any player could challenge the machine, the win rate was 99.4 percent. This confirmed the ability of the computer to cope with tasks in conditions of limited information.
Further development of artificial intelligence, according to experts, can lead to a revolution in military affairs, comparable to the invention of gunpowder and nuclear weapons. However, another analogy seems much more appropriate — mastering fire. This allowed man to become the king of nature. There are now fears that people will be dethroned by a tool of their own design if given unlimited access to weapons.