In what could sound like the death knell for human pilots in future warfare, China’s infra-red AI has managed to outmaneuver ace military aviators who rely on their skills and unpredictable, high-intensity aerial acrobatics.
Earlier in 2020, Pentagon researchers pitted a man against AI in an aerial dogfighting tournament to see if artificial intelligence algorithms could outsmart F-16 pilots.
The AI algorithm, developed by Heron Systems, pummeled an F-16 pilot with 2000 hours of flying experience in a simulated dogfight five times. These were scripted simulations. However, AI-driven aircraft took part in real-world dogfights only in 2024.
While most of the world’s air forces are already researching Combat Collaborative Systems, the world was hopeful that human pilots could prevail over AI in aerial battles, relying on their unpredictability. However, a recent Chinese study has snuffed out the last hope for human pilots.
The technique, which was described in research released at the end of 2024, uses AI-driven predictive modeling in conjunction with sophisticated infrared vision to identify minute wing-tail motions and predict an opponent’s next move.
According to a group of scientists from the Northwest Institute of Mechanical & Electrical Engineering, located in the northwest city of Xianyang and a major research arm of Norinco, China’s largest arms supplier, the development could make even the nimblest fighter jets, like the US-made F-15, essentially defenseless. The study was published in the December issue of the Journal of Gun Launch & Control.
So far, the AI air combat systems have relied on trajectories to forecast the fighter jets’ next move. This made it difficult to account for human pilots’ abrupt, non-linear maneuvers. However, by concentrating on the physical mechanics of enemy aircraft, the Chinese team, under the direction of senior engineer Lin Zhiwei, was able to circumvent this restriction.
“Human pilots rely on instinct and unpredictability, but every physical maneuver has mechanical precursors,” Lin and his team wrote in the paper.
The system analyzes infrared footage using a modified YOLOv8 neural network to identify millimeter-level deformations in an opponent’s control surfaces during flight, such as the F-15’s 1.5-meter (five-foot) rudder or two-meter (6.5-foot) elevator.
A long short-term memory (LSTM) network enhanced with attention-weighing mechanisms receives these real-time observations, allowing the AI to anticipate movements before they fully materialize.
The new prediction method could reduce the targeting errors to under two meters. This is a tenfold improvement over the traditional prediction methods, making it difficult for human pilots to outfox the AI algorithm. The AI can guide the shells to hit the cockpit if necessary.
The ‘Red Eye’ AI was tested against the highly complex aerial profiles of F-15s in the real world, and every time, it was able to be one step ahead of the human pilot.
AI-Driven 7th Military Revolution
‘The fighter jet era has passed.’ These words were famously spoken by Elon Musk at the 2020 Air Warfare Symposium when prompted to describe the future Air Domain.
AI has emerged as another frontier for competition between China and the US. China has swiftly revolutionized its military capabilities by investing heavily in artificial intelligence and utilizing aggressive covert and overt technological appropriation. China has dedicated its efforts to dominate commercial and civilian AI systems.
At the moment, the US is leading in terms of AI applications in the military, but China is not far behind.
The US Department of Defense is spending billions of dollars to create and incorporate AI into its defensive systems. In the 2024 defense budget bill, the DoD asked for US$1.8 billion for AI and machine learning. To establish an AI network of interconnected sensors across all branches of the military, the DoD has asked for US$1.4 billion for its Joint All-Domain Command and Control Initiatives.
The efforts have also been bearing fruits, as the AI-controlled F-16, Vista, went up against another human-piloted aircraft in 2024. This marked one of the biggest advancements in military aviation since the introduction of stealth technology, a giant leap that could make human pilots redundant.
Former USAF Secretary Frank Kendall demonstrated the reliability of the AI-piloted aircraft while sitting in the cockpit of the experimental F-16 jet, X-62A, or VISTA (variable in-flight simulator test aircraft), which flew at a lightning-fast speed of over 550mph over Edwards Air Force Base.
The experiment went up against the human-piloted F-16 as both came within 1000 feet of each other, twisting and looping to force their opponent into a defensive position.
The engineers working on Vista want the first fleet to be ready by 2028 and say the programs are learning so quickly that some are already beating human pilots in combat. The US Air Force is aggressively pursuing AI integration, aiming for a fleet of over 1,000 unmanned warplanes by 2028.
Vista’s military operators contend that it is the first of its kind AI aircraft in the world. The software first learns from millions of data points in a simulator and then tests its conclusions during actual flights. The real-world performance data is fed into the simulator, and the AI processes it to imbibe the learnings.
