Limiting AI use in military applications and ensuring global AI governance could ease collaboration and dialogue between China and the US as they vie for influence over the transformative technology, according to experts at a forum in Hong Kong on Monday.
Their remarks came as artificial intelligence is being increasingly adopted in the defence sector, including for weapon-related functions. It raises serious ethical and accountability concerns, further elevates the US-China AI race to a high-stakes geopolitical contest and adds to existing tensions over regulation standards, talent and access.
Sun Chenghao, a fellow at Tsinghua University’s Centre for International Security and Strategy, said AI regulation in the military domain and global governance that aimed to address universal challenges could provide common ground for China and the US to foster bilateral cooperation.
Speaking on the sidelines of the US-China Hong Kong Forum, Sun said the active engagement of major powers in governance was the basis for Global South countries to take part, but that tensions between China and the US, who led AI development, made effective AI governance cooperation “logically difficult”.
At the annual World Artificial Intelligence Conference in Shanghai in July, China proposed a global AI cooperation organisation be created and headquartered in Shanghai, marking a further step in advocacy for AI inclusiveness and equal access to the technology for the Global South.
During the forum, Christopher Nixon Cox, a board member of the Richard Nixon Foundation, told a panel discussion on the ethical and equitable application of emerging technologies that bioweapons were an “obvious area” where the US and China should work together to “limit the influence of AI”.
“There is no good that would come out of having AI go off and develop bioweapons that are easy to replicate, or you can spread around the world,” he said.
Cox also called for both sides to start by having a conversation about leading AI regulation.
His view was echoed by Zhang Tuosheng, an academic committee member at Tsinghua University’s Centre for International Security and Strategy (CISS), who urged Beijing and Washington to resume intergovernmental dialogue on AI as soon as possible and consider making military AI application an important component of the conversation.
Beijing and Washington held their first inter-governmental dialogue on AI in May last year, a meeting that covered the risks of the technology and ways to manage them. A second meeting has yet to take place.
Zhang said the two sides should implement the major consensus reached by their leaders last year.
In November last year, during their meeting in Lima, Peru, Chinese President Xi Jinping and his then-counterpart Joe Biden agreed that humans, not artificial intelligence, should decide the use of nuclear weapons, a first-of-its-kind step on the issue.
But, earlier in September 2024, China had refused to sign a global legally non-binding pact outlining the responsible use of AI in the military at a global summit in Seoul, a move that embodied the stark difference in views among the stakeholders.
Sun Chenghao, of CISS, said intense competition had arisen from a focus on urgent geopolitical issues such as export controls, while the potential risks posed by the evolving AI seemed less immediate in this landscape.
“It is hard to imagine the Chinese government deciding to cooperate closely with the US on governance while facing US restrictions on hardware. Logically, that just doesn’t make sense,” Sun said.
Washington has imposed export restrictions on advanced technology to China, including the most high-end AI chips and chipmaking equipment, citing concerns that the Chinese military would use the technology to increase its capabilities.
But as China steadily advanced its technology and narrowed the chip gap, prompting some in the US to rethink the export restrictions, Sun said he held a positive outlook for the competition.
Amid the US tech curbs, China has accelerated the development of technology and reached breakthroughs. In August, it launched an ambitious “AI+” initiative to upgrade diverse sectors and spur new products and services.
On the civilian front, Lee Kai-Fu, founder and CEO of Chinese start-up 01.AI, told the forum that American companies were advised to consider the “Chinese approach” in their AI development, which better reflected reality and could make US-China competition more constructive.
“The US-China competition would be better and more constructive if more people understood, and perhaps embraced, the Chinese approach,” said keynote speaker Lee, the former head of Google China.
He said that while the US, shaped by Silicon Valley’s winner-take-all mindset, followed a “giant-step function” in building artificial general intelligence (AGI), Chinese companies pursued a practical approach focusing on incremental progress, open-source sharing and user-driven competition, rather than trying to build a world-dominant platform.
As AI has advanced rapidly, AGI has emerged as the holy grail for developers. It is a theoretical AI system with capabilities that rival those of a human. This contrasts with narrow AI, which is designed for specific tasks.
“As the US spends trillions of dollars dreaming or hoping to build that one giant AGI model to squash everyone else, China is sort of like collaborating and building open source, and trying to figure out ways to make money,” Lee said.
Silicon Valley thinking naturally caused the US government to believe that “as one company squashes other companies, one country will squash other countries” but that assumption “may not be correct”, he said.
“If you look at the progress of AI for the last year, the Chinese visualisation is more accurate ... There is no giant-step function, at least looking historically,” he said, adding that the Chinese approach was not governmental, as it followed a “naturally adopted” leader-and-follower dynamic.
“This is not America’s bad [and] China’s good. The leader naturally wants to advance and become No 1 ... But the follower usually says, OK, let’s be open-source. Let’s see how we can do to get more adopters.”
Even if one company achieved the breakthrough a few months earlier, others would quickly catch up because they were “not that far apart”, Lee said. -- SOUTH CHINA MORNING POST
