Killing machine. Artificial intelligence defeats US Air Force pilot in virtual combat
The US Department of Defense held a virtual AI battle with an F-16 fighter pilot. The algorithm protested the maximum capabilities of man and aircraft.
Recently, the Office of Advanced Research Projects of the US Department of Defense (DARPA) announced a virtual battle between artificial intelligence and an Air Force F-16 fighter pilot.
The algorithm was developed by the Californian company Heron Systems, and before a fight with a real Air Force pilot, a computer program defeated seven other challengers.
The person who took part in the virtual battle is a DC National Guard pilot, an active fighter pilot with over 2000 hours of F-16 experience.
In a simulation of five different basic scenarios for maneuvering a fighter in combat, Heron Systems’ AI defeated a man and was able to destroy his plane.
At the same time, the algorithm specified the limits of the F-16’s capabilities and the permissible overload values for this aircraft. An Air Force pilot said the program maneuvers and does not act the way humans do.
The authors of the program say their AI was guided by the habits and rules of training for Air Force pilots. For example, instructions for American pilots prohibit getting too close to another aircraft and limit the angle of attack when firing a cannon. The algorithm did not have to follow these rules, which allowed it to gain an advantage.
The purpose of this test is to test the possibilities of using AI in combat aircraft. Heron Systems’ algorithm triumphed over other similar AI systems under development at Aurora Flight Sciences, EpiSys Science, Georgia Tech Research Institute, Lockheed Martin, Perspecta Labs, PhysicsAI, and SoarTech.
According to the head of the DARPA air combat development program, Colonel Daniel Javorsek, even if today the AI system was perfect and ready for operation, it would take about 10 years to introduce it into an F-15 or F-16 combat fighter.
Source: nv.ua
