Once Angering the U.S. Over a Nuclear-Weapons Supercomputer, China’s National Defense University Unveils AI Course Built Around Combat
Translated from Chinese and summarized by DistantNews. Read the original for the full story.
At a glance
- Researchers at China’s National University of Defense Technology proposed redesigning a graduate AI course around military decision-making and combat scenarios.
- The proposed curriculum would cover battlefield assessment, decision-making, targeting and battle-damage assessment through exercises involving simulations, autonomous systems and wargaming.
- The researchers said AI should remain a decision-support tool, with humans retaining full control over command and the use of force.
China’s military researchers want graduate students to learn artificial intelligence through the demands of war. Their proposal centers on an AI course guided by the principle of “education for war,” with actual combat scenarios woven into the curriculum.
The idea came from a research team at the Intelligent Science Academy of the People’s Liberation Army’s National University of Defense Technology. The team presented its proposal in the July issue of Defence Industry Conversion, after reviewing a course first offered in the autumn semester of 2022 to science and engineering graduate students without AI backgrounds.
That course, “Principles and Practice of Artificial Intelligence,” aimed to help students connect emerging technology with their own fields. But the researchers concluded that its case studies lacked enough military content. They proposed rebuilding the course around common unmanned-operation scenarios and using military operations to link the different AI theories.
The planned subjects would correspond to stages of military decision-making and strike operations, from observing and assessing a battlefield to making decisions, taking action, locating and tracking targets, engaging them and evaluating the results. Reinforcement learning would be taught through combat decisions and wargames. Uncertainty reasoning, which addresses incomplete or unreliable information, could be practiced through fault diagnosis for unmanned systems and border-intrusion warning systems.
The researchers said students should not merely run existing code. They should receive defined tactical problems, datasets, algorithms and step-by-step tasks. One proposed exercise would have students train several simulated drones to operate together, surround and lock onto a high-value target. An online simulation platform would let teams design maps and combat scenarios, assign forces and missions, train AI agents through simulated battles, and compete against one another.
The university, which reports directly to China’s Central Military Commission, is one of the country’s leading military technology institutions. The U.S. Commerce Department placed it on an entity list in 2015 over alleged use of U.S. technology in a supercomputer connected to nuclear testing activities. Despite the course’s focus on military applications, its authors set a firm ethical boundary: AI may assist decisions, but humans must retain complete control over command and the authority to open fire.
education for war
Originally published by Liberty Times in Chinese. Translated, summarized, and contextualized automatically by DistantNews, with a note on how the source frames the story. Not individually reviewed before publishing. How this works.