Beijing, Washington to hold first high-level talks on ‘risks’
The United States and China will hold their first high-level talks on artificial intelligence (AI) in Geneva today, according to senior US officials.
The discussions will focus on the emerging risks associated with AI systems, and how the two countries define “risk” and “safety”.
Both sides will explain their domestic approaches to addressing the risks and to setting norms and principles on AI safety, as well as exchanging views on international governance, according to one official.
“[The talks will] not focus on any particular deliverables, but rather an exchange of views on the technical risks of AI,” he said, adding that no joint statement was expected afterwards.
The official also said the meeting would not focus on “promoting any form of technical collaboration or cooperating on frontier research in any matter”.
The talks follow an agreement in November between Presidents Xi Jinping and Joe Biden about the need to address the risks of advanced AI systems and to improve AI safety through talks. The two sides have since had several conversations on setting an agenda for the coming talks.
Beijing and Washington have differing priorities in the bilateral dialogue about AI, according to observers.
“For Washington, the primary focus is the management of the development of frontier AI models, high-risk ones particularly,” the Institute for China-America Studies, a Washington-based think tank, said in a note last month.
Beijing would “prefer the bilateral conversation be focused on the cooperative development of AI”, the note said.
But even if the US does not want to cooperate on developing AI, China believes it can gain insight and influence through international discussions, and this is a key motive for its involvement in them, according to Tong Zhao, a senior fellow at the Carnegie Endowment for International Peace.
Another motive “is to demonstrate China’s image as a responsible power who is willing to play a leadership role in key areas of international governance”, he said.
Today’s meeting will be attended by representatives from the Ministry of Foreign Affairs and the National Development and Reform Commission.
Tarun Chhabra, senior director for technology and national security at the National Security Council, and Seth Center, acting special envoy for critical and emerging technology at the State Department, will lead the US delegation.
During the talks, the US side plans to express concern over the security risks from China’s use of AI.
“[Beijing] is rapidly deploying capabilities across civilian as well as military, national security sectors, and in many cases in ways that we believe undermines both US and allied national security,” the official said.
Washington has in recent years tried to choke off China’s access to AI technology. In October 2022, the White House introduced export controls restricting leading US designers, such as Nvidia and AMD, from selling their high-end chips for AI and supercomputing to China.
But while technology protection policies were “not up for negotiation”, Washington was open to starting a channel to discuss these issues, the US official said.
A second official said: “Our credibility with reaching much of the world with what we believe is the right approach to safe, secure and trustworthy AI will be enhanced if we’re seen as engaging seriously with the PRC [People’s Republic of China] on safety and risk issues.”
Both the US and China have been working towards creating national standards for AI regulation.
In July, China issued the world’s most detailed regulations on generative AI models, while the White House issued its first executive order on the development and use of AI in October.
There have also been growing signs that the two are willing to cooperate on regulating AI. Last November, both countries were among the two dozen or so participants that signed an agreement affirming the need to address risks that could arise from AI at a summit in Britain.
In March, China co-sponsored, along with more than 120 countries, a US-led, non-binding United Nations General Assembly resolution on artificial intelligence – the first of its kind.
But China did not join a US-led resolution on the responsible military use of artificial intelligence in November.
While observers said Washington and Beijing would both be in favour of requiring human input in decision systems for autonomous weapons and nuclear warheads, the two sides have not yet come to a public agreement on the matter.
“China’s policy community supports keeping humans in the loop and limiting the use of AI in nuclear weapon systems,” Zhao said, but the topic was “politically unappealing” to the leadership in Beijing.
He also said he did not expect the talks to delve substantively into military uses of AI.
[The talks will be] an exchange of views on the technical risks of AI A SENIOR U.S. OFFICIAL