AI’s Ascent: The End of the Line for Human Pilots?
China’s latest advancements in artificial intelligence, specifically its infra-red AI capabilities, may signal a significant shift in the landscape of future warfare. This AI has demonstrated the ability to outmaneuver even the most skilled military aviators, who traditionally rely on their expertise and unpredictable, high-intensity aerial maneuvers. This development raises serious questions about the necessity of human pilots in aerial combat.
Early successes of AI in simulated dogfights had already hinted at this potential. In 2020, researchers from the Pentagon conducted a trial where an AI algorithm, developed by Heron Systems, engaged in a simulated aerial dogfighting tournament against a seasoned F-16 pilot with thousands of hours of flight experience. The AI algorithm won convincingly, defeating the human pilot five times. While these results were initially based on scripted simulations, subsequent real-world dogfights in 2024 further solidified the AI’s capabilities.
While most of the world’s air forces are actively researching Combat Collaborative Systems, there was a prevailing hope that human pilots could maintain an edge in aerial combat due to their inherent unpredictability. However, a recent study from China has cast doubt on this belief. The research, which was released at the end of 2024, details a sophisticated system that leverages AI-driven predictive modeling. This system works in conjunction with advanced infrared vision to identify minute wing-tail motions and predict an opponent’s next move. The study was published in the December issue of the Journal of Gun Launch & Control.
This new system represents a significant leap forward, particularly when compared to earlier AI-driven air combat systems. These older systems typically relied on predicting the trajectory of fighter jets, which made it difficult to account for the abrupt, non-linear maneuvers of human pilots. By focusing on the physical mechanics of enemy aircraft, the Chinese team was able to circumvent this restriction. Senior engineer Lin Zhiwei, who led the research, noted that, while human pilots rely on instinct and unpredictability, every physical maneuver has mechanical precursors. The system analyzes infrared footage using a modified YOLOv8 neural network, identifying even the smallest changes in control surfaces during a flight – such as the F-15’s rudder or elevator.
These real-time observations are then processed by a long short-term memory (LSTM) network, enhanced with attention-weighing mechanisms. This setup allows the AI to anticipate movements before they fully materialize, potentially reducing targeting errors to under two meters. This represents a tenfold improvement over previously used prediction methods, making it exceedingly difficult for human pilots to outmaneuver the AI algorithm. During testing against the complex aerial profiles of F-15s, the AI was consistently one step ahead of the human pilot.
The Future of Aerial Warfare
This technological advancement arrives amid a period of rapid innovation in military AI. Elon Musk famously stated at the 2020 Air Warfare Symposium that the fighter jet era had passed, and AI has emerged as a new frontier for competition between the US and China. China has been swiftly modernizing its military capabilities by investing heavily in AI and utilizing technology from various sources. While the US currently leads in military AI applications, China is rapidly closing the gap.
The US Department of Defense is spending billions to create and incorporate AI into its defensive systems. The 2024 defense budget included US$1.8 billion for AI and machine learning, and an additional US$1.4 billion for Joint All-Domain Command and Control Initiatives. The US efforts also include an AI-controlled F-16, known as Vista (variable in-flight simulator test aircraft).
The US Air Force is aggressively pursuing AI integration, with aspirations of a fleet of over 1,000 unmanned warplanes by 2028. Engineers working on the VISTA program aim to have the first fleet ready by 2028, claiming that these programs learn so rapidly they are already outperforming human pilots in combat scenarios.
The implications of this emerging technology are profound, potentially revolutionizing aerial combat. With the capacity to outmaneuver human pilots, AI could reshape the future of military aviation, making human pilots increasingly redundant.
