Chinese military AI attack systems lag foreign models, researchers say

China’s military AI systems remain more useful for answering questions and generating content than for attacking targets, according to a study by Chinese researchers.

The paper says military applications are concentrated in auxiliary functions and still perform considerably worse than foreign large language models. It adds that offensive uses in actual combat remain at an exploratory stage, supported by relatively limited theoretical and empirical research.

“

The military’s AI applications are mainly used in auxiliary fields such as intelligent question-answering and content generation.

The findings appeared in July in the journal Defence Industry Conversion in China. The authors are affiliated with the Armed Police Force Shandong Corps and the Engineering University of the People’s Armed Police.

The assessment comes as artificial intelligence shifts from an emerging military technology to a core element of strategic competition between the United States and China. Both countries’ militaries are examining how AI could improve command and control, but the study portrays China’s current military models as lagging foreign equivalents and not yet ready for extensive combat applications.

“

They still lag considerably behind foreign large language models in performance, while offensive applications in actual combat remain at an exploratory stage, with relatively limited theoretical and empirical research.