US-China distrust hinders cooperation on AI safety
Mutual distrust between the United States and China may be one of the main risks to the development of artificial intelligence, as rivalry between the two countries hinders coordination in technology safety. CBS News reports, citing researchers and AI regulation experts.
Experts warn about the risks of the emergence of superintelligence — a theoretical AI system smarter than any human. They fear that such a system could potentially view humans as competitors for necessary resources, creating a threat to humanity’s survival.
Competition instead of coordination
Executives of the US companies OpenAI and Anthropic have called for slowing the development of advanced AI systems. At the same time, Chinese leaders have also spoken about the need to control the technology. In June, Chinese President Xi Jinping warned of the danger of losing control over AI, and in July said that countries must cooperate to ensure the technology always remains under human control.
Leah Wang, a researcher at the Carnegie Endowment for International Peace, said that Chinese companies generally share the increasingly widespread view that AI needs regulation. She described the Shanghai Artificial Intelligence Laboratory as a reputable institution working on the safety of this technology.
According to Wang, technological competition is the top priority for US AI leaders when it comes to China. Chinese authorities, in turn, view calls to slow development as an attempt by the United States to preserve its technological dominance. China’s Foreign Ministry said that threat narratives only hinder the process of global AI governance.
More current news is available on the UA.News Telegram channel Telegram.
Proposals for dialogue
Nvidia CEO Jensen Huang previously said that US and Chinese researchers should maintain contact. In his view, there is a lack of bilateral coordination among industry representatives because China is treated as an adversary.
Kwan Yi Ng, head of international governance at the Beijing consulting company Concordia AI, believes that a large-scale international agreement or coordinated slowdown in development may not be the best format for engagement with Chinese companies and the government. She proposed starting with an exchange of cybersecurity practices, coordination to counter the use of AI by terrorist groups, and technical contacts between lower-level government officials.
Carnegie Endowment researcher Matt Sheehan proposed an approach of “parallel AI safety”: the United States and China could develop technologies along separate paths while reducing risks that each side considers important to its own national security.
After the May meeting between Donald Trump and Xi Jinping in Beijing, the Chinese side said that the leaders had agreed on an intergovernmental dialogue on AI. Trump confirmed that he had discussed with Xi the idea of safeguards for the development of the technology. The next meeting of the presidents in Washington is scheduled for September 24, and AI safety is expected to be among the topics of the talks.
