China and U.S. explore common ground on AI governance
China and the United States may not be able to resolve all their differences over artificial intelligence at once, but they can begin by managing the most urgent risks and gradually expand areas of common understanding, according to a commentary published by Yuyuantantian, a media arm of China Media Group.
AI governance has been discussed between representatives of the two countries through several channels this year, including the United Nations, G20, APEC and expert-level Track II dialogue. In late September, China and the United States also agreed to establish a communication channel for AI-related incidents.
Experts from both countries are discussing the creation of an incident-reporting mechanism that would allow AI laboratories and government agencies to exchange information about potentially dangerous developments. Preventing the misuse of advanced AI capabilities by terrorist organizations and other malicious actors has also emerged as an area for possible cooperation.
Differences remain, however, in how Chinese and U.S. researchers assess AI risks. U.S. experts have placed greater emphasis on scenarios involving large-scale loss of control over AI systems, while Chinese researchers tend to focus more on specific risks arising from the technology's practical use.
Issues requiring further discussion include common standards for classifying AI risks, definitions of serious model loss-of-control events, and rules and procedures for reporting AI-related incidents.
The commentary also argues that AI governance should be discussed within a broader international framework involving more countries, as technologies such as advanced models and deepfakes can have consequences for public safety and social trust.
According to CCTV+, despite continued competition in artificial intelligence, China and the United States have opportunities to cooperate on AI safety by first addressing the most serious risks and establishing practical communication mechanisms.
