No Chinese AI impact --- too many barriers to development

Second, that said, Chinese defense experts are concerned that the PLA is lagging behind the U.S. military in AI and related emerging technology development and deployment.

Whereas U.S. experts increasingly voice concern that China has overtaken the United States in the development and adoption of military applications of AI and related emerging technologies, the articles reviewed in this paper reveal an inverse anxiety across the Pacific; Chinese scholars believe the United States is ahead in the race to develop and field AI-enabled military systems.170 Third, Chinese defense experts identify certain challenges in developing and fielding military AI systems, which are similar to those the U.S. military appears to be facing.

The U.S. military’s AI development efforts are not the focus of this paper. Moreover, the report does not provide systematic comparisons of U.S. and Chinese experts’ views of each country’s AI-related challenges. That said, based on publicly available assessments of the U.S. military’s AI development and adoption efforts and statements made by U.S. military officials, both countries do seem to be dealing with similar sets of issues related to developing and deploying the AI and related emerging capabilities detailed in this paper.171 They seem to be facing challenges related to creating trustworthy military AI, as well as to managing and fully exploiting data resources while preserving the integrity of communications and computer networks. Fourth, the authors also identify challenges that may be especially relevant to China and the PLA regarding the development and adoption of military AI and related emerging technologies.

Specifically, the Chinese observers surveyed in this paper note difficulties collecting, managing, and using military-relevant data to train military AI systems, as well as developing the high-quality sensors experts believe will be needed to fight nextgeneration wars. Some scholars have also noted concerns about the state of China’s communications and cybersecurity systems, and their ability to connect vast arrays of sensors and process the information they generate. These observed shortcomings and obstacles raise important questions about China’s progress toward implementing its systems warfare concept, which is a central aspect of China’s theory of victory in future wars.172

The authors’ views on China’s apparent deficiencies related to establishing standards and T&E protocols also merit mention. A lack of military standards complicates the development of interoperable systems, while poor T&E practices create obstacles to trusting AI-enabled systems, determining their efficacy, and training them for future wars in complex environments. According to these papers, the lack of robust standards and T&E resources are some of the most serious threats China faces in creating the systems that would allow it to fight future wars according to its nascent operational concepts.173

Fifth, Chinese defense scholars voice concerns about AI risks and the trustworthiness of AI systems.

Chinese scholars’ aforementioned concerns about the explainability, robustness, controllability, and overall trustworthiness of AI systems reveal doubts about the current state of the technology. Potentially more significant, however, are the scholars’ misgivings about using AI technologies that are not explainable, predictable, or controllable. They explain that such deficiencies could lead to unwanted outbreaks and escalations of wars, as well as to civilian and friendly force targeting by autonomous systems. Moreover, they note that insufficiently trustworthy systems may be less effective for military operations overall, as such shortcomings will inevitably threaten operator trust, making soldiers less likely to use them effectively (or at all).

In short, the majority of experts who discuss the shortcomings of AI systems question whether the technology can be used responsibly and effectively, at least in its current state. Notably, these doubts are largely in line with the Department of Defense’s own concerns about using AI systems that are insufficiently trustworthy in operational settings.