President Biden’s planned work with China on artificial intelligence rules is facing pressure from critics warning that the communist government cannot be trusted and from allies looking to steer future cooperation.
After meeting with Chinese President Xi Jinping on Wednesday, Mr. Biden said the two nations will arrange meetings with AI experts to discuss risk and safety issues.
“The United States will continue to compete vigorously with the [People’s Republic of China] but will manage that competition responsibly so it doesn’t veer into conflict or accidental conflict,” Mr. Biden told reporters. “And where it’s possible, where our interests … coincide, we’re going to work together like we did on fentanyl. That’s what the world expects of us.”
Details on forthcoming AI meetings are scant, but critics say the Biden administration must not trust China to live up to any deal it may make.
Nathan Beauchamp-Mustafaga, a researcher at the RAND Corp. think tank, said the Chinese government is a liar and the U.S. should look to call its bluff.
“Biden should make clear to Xi, publicly and privately, that any Chinese interference [via AI] in the upcoming 2024 elections will be unacceptable and met with a strong response,” he said on RAND’s blog.
SEE ALSO: Nikki Haley calls Biden’s meeting with China’s Xi ‘embarrassing’
Mr. Biden did not elaborate on the AI risks that China and the U.S. will address in future discussions, but some policy experts think nuclear armageddon is at the top of the list.
Before the rendezvous with Mr. Xi in San Francisco, rumors swirled that Mr. Biden would announce a deal with China to ban the use of AI in weaponry, including nuclear warheads and drones. Instead, the U.S. president’s team said the two leaders only identified a need to address AI risks and shared a goal of improving AI safety.
With no AI deal between the U.S. and China, some of Mr. Biden’s Democratic allies in Congress want America to adopt a new law to keep a human in the decision-making loop for nuclear warfare.
Sen. Edward Markey, Massachusetts Democrat, and Rep. Ted Lieu, California Democrat, are leading a push to pass the Block Nuclear Launch by Autonomous Artificial Intelligence Act, which has the support of four Republican representatives.
The Democratic duo joined the Future of Life Institute on Capitol Hill to screen a short film imagining a spiraling conflict between the U.S. and China sparked by faulty AI systems.
After the broadcast, Mr. Markey on Wednesday said new laws are needed regardless of whether the U.S. and China reach any agreement on preventing AI from use in autonomous weapons systems.
“Nothing that President Xi says to President Biden today is binding on the next two presidents, which is why you need laws,” Mr. Markey said at the event. “You need laws enshrined to make sure that those individuals who succeed those people who just gave their word are maintaining past policy in the future.”
Changes in House leadership and Senate debates over other tech show that action on new rules for AI is not lawmakers’ most urgent priority.
Mr. Biden issued a sweeping executive order last month seeking to limit the potential harm from AI and shape the U.S. government’s use of it. The order pushes AI developers to share results from testing the tech with the feds, and the order tasks government officials with making new requirements for AI makers.
The State Department on Monday said 45 countries are working with the U.S. to implement new measures to guide the responsible use of AI and autonomy for military applications.
China has not joined those countries’ Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy. Instead, the Chinese are drafting rules for generative AI amid the growth of new tech tools creating content in response to users’ queries.
Last week, Georgetown University’s Center for Security and Emerging Technology published an English-language translation of China‘s draft standard that indicates the AI rules the communist government wants to impose.
Georgetown’s Alex Friedland wrote in the center’s policy.ai newsletter that the draft policy proposes oversight processes that Chinese AI companies must incorporate for their data and content.
“The standard names more than 30 specific safety risks, some of which — algorithmic bias, disclosure of personally identifiable information, copyright infringement — are widely recognized internationally,” Mr. Friedland wrote Thursday. “Others, such as guidelines on how to answer questions about China‘s political system and Chinese history, are specific to the tightly censored Chinese internet.”
Regardless of whether China makes any deal with the U.S. for AI rules, American officials aren’t optimistic they can constrain adversaries’ use of new software and AI tech.
Ambassador Nathaniel C. Fick, inaugural leader of the State Department’s cyberspace bureau, told lawmakers this week that stopping enemies from using software and AI tools will prove more difficult than stopping nuclear proliferation.