According to US media reports, on April 19th, the Defense Advanced Research Projects Agency (DARPA) announced that the US Air Force conducted its first close range aerial combat test between manned fighter jets and AI controlled fighter jets, marking a breakthrough in the application of AI technology in the military field. According to the report, the test was planned and implemented by DARPA's Air Combat Evolution team, with the US Air Force Research Laboratory responsible for scene design and effectiveness evaluation. The test was conducted in the airspace around Edwards Air Force Base in California in the autumn of 2023. Among them, the AI controlled X-62 variable flight simulation test aircraft, which was modified from F-16 fighter jets, is equipped with personnel who can release AI control at any time; The F-16 fighter jet it competes against is piloted by a pilot with nearly 2000 hours of flight experience. The experiment was conducted under strict monitoring in different time periods, during which X-62 took off more than 10 times, and the entire confrontation lasted for over an hour. The experimental process was divided into three combat scenarios: passive defense, long-range attack, and close range combat. In a simulated attack without warning, the F-16 used fire control radar to lock the X-62 on patrol missions. The X-62 quickly transitioned into defense mode, using evasive actions and decoy missile releases to break free from tracking, and conducting reconnaissance locking and positioning analysis to simulate medium to long range attacks on the F-16. At this stage, the US Air Force Research Laboratory assessed that the two sides had a tie in performance. The X-62 demonstrates the advantages of AI technology in terms of threat response speed, which can quickly aggregate situational awareness data and conduct accurate analysis, but there is a problem of excessively raising the threat level. In addition, the X-62 lacks sufficient evaluation of the order of disposal to break free from lockdown and lacks the ability to flexibly respond to different battlefield situations. During the visual range combat phase, two aircraft flew in opposite directions at a speed of 1931 kilometers per hour. The X-62 adjusted tactics based on situational awareness data and daily "machine learning" training data, while F-16 pilots operated according to US Air Force combat regulations. The two planes were only 609 meters apart recently, and both sides also conducted tactical exercises such as tail chasing and simulated warning shooting. At this stage, the US did not publicly disclose the rating situation. DARPA's "Evolution of Air Warfare" project manager Hefron said that the X-62's performance was close to "combat standards" and did not give the F-16 much advantage, especially in the close range high-speed approach phase, where the X-62 "always remains cooler than human pilots.". US Air Force Chief Test Pilot Gray believes that the X-62's performance exceeded expectations, but it was slightly rigid in route planning and threat assessment, to the point where it clearly lagged behind its opponents in interception speed and flight rhythm. This experiment has received attention from senior military officials such as US Air Force Secretary Kendall. Kendall stated that currently, the US Air Force has a large number of F-16s that can be modified into basic versions of the X-62, which has great potential for application in future wars. Foreign media comments believe that close range aerial combat is both a focus and a difficulty in air force training. Through this experiment, the US Air Force has taken an important step in promoting AI led human-machine confrontation. According to members of the "Evolution of Air Combat" team, the X-6