The 11th Beijing Military Intelligent Technology Expo has opened a window into the People’s Liberation Army’s accelerating pivot toward 'intelligentized' warfare. No longer relegated to back-end logistics, artificial intelligence (AI) has moved to the tactical edge, manifesting as 'AI pilots,' multi-modal target recognition models, and high-speed interceptor drones. This shift reflects a strategic recognition that the sheer volume of battlefield data now exceeds human cognitive capacity.
Central to this display was the 'Junzhi' Military Large Model, a generative AI framework designed to process massive streams of satellite and drone reconnaissance. Unlike traditional computer vision, this model can identify not just the presence of a vehicle, but its operational state—such as a rocket launcher transitioning to a firing position. By training on hundreds of thousands of data points, these systems are designed to overcome the persistent challenge of false positives in complex terrain.
To refine these algorithms, Chinese defense firms are bridging the gap between digital training and physical reality through high-fidelity, 1:1 scale targets. These robotic decoys mimic the visible, infrared, and radar signatures of main battle tanks and mobile artillery, providing a 'ground truth' for AI training. This systematic approach suggests China is focusing heavily on the data-labeling and validation bottlenecks that currently plague global military AI development.
In the aerial domain, the focus has shifted toward manned-unmanned teaming through a 'Three-Member' AI architecture. By deploying AI pilots, commanders, and navigators alongside human crews, the system aims to offload decision-making stress during high-speed, beyond-visual-range engagements. While the final authority to fire remains with the human pilot, the AI manages tactical maneuvers and target allocation, effectively acting as a cognitive exoskeleton.
Counter-drone technology emerged as another dominant theme, driven by the clear lessons of recent regional conflicts. Exhibitors showcased a new generation of high-speed interceptor drones capable of reaching 400 kilometers per hour. These 'kamikaze' interceptors are designed to fill the defensive gap between small arms and expensive surface-to-air missiles, utilizing AI-driven 'fire-and-forget' visual tracking to negate electronic jamming.
Ground and maritime autonomy also displayed a trend toward interoperability and swarm coordination. The introduction of 'heterogeneous unmanned system controllers' allows robot dogs from different manufacturers to communicate and execute coordinated strikes alongside aerial drones. This emphasis on a unified 'brain' for disparate systems indicates a move toward a more resilient, integrated kill web that can function even when individual nodes are lost.