US–China AI Competition At The Crossroads Between Dialogue And Decoupling – Analysis

By Emmie Hine

The United States and China are competing fiercely in the race to develop artificial intelligence (AI). While Beijing is working to lead the world by 2030, Washington’s approach and US–China competition hinge on the result of the 2024 US presidential election between Democrat Kamala Harris and Republican Donald Trump.

China is pushing ahead with both its AI development and regulation. Despite discouraging headwinds from US chip sanctions and concerns that regulatory requirements may stifle innovation, the Chinese government has proved willing to bend its rules to promote development and social stability. China’s AI progress will likely continue. Though behind in many metrics, it holds the majority of global AI patents, and its large language models starting to top some international leaderboards.

The US relies heavily on self-regulation and voluntary compliance to regulate the industry. The Biden administration’s AI governance approach has been light on legislation that impacts the private sector. But it has laid out detailed directions to federal agencies for best practices and developing guidelines, forwarding a community-centred and rights-based narrative.

Washington’s flagship AI governance document — the October 2023 Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence — takes a whole-of-government approach towards AI development. It aims to promote AI use across the federal government, encourage innovation, protect civil rights and strengthen US leadership on the international stage. Though Harris may support more concrete AI regulation, any legislation would require overcoming a congressional gridlock. A future Harris administration would likely continue Biden’s relatively light-touch approach to restraining the AI industry.

The only private sector requirements in Biden’s executive order require developers of frontier models and companies with large computing clusters to make reports to the federal government. But it has still faced criticism from some in Silicon Valley.

While many AI startups and Big Tech companies alike approved of the AI executive order, the 2024 Republican Party platform has gained support from prominent figures who reject any attempt to regulate AI, including Marc Andreessen and Elon Musk. The platform pledges to repeal the ‘dangerous’ executive order in favour of ‘AI Development rooted in Free Speech and Human Flourishing’ and claims that Biden’s directive ‘hinders AI Innovation’ and ‘imposes radical leftwing ideas’ on its development.

Trump allies have drafted a new executive order to ‘Make America First in AI’ through military technology moonshots and eliminating ‘unnecessary and burdensome regulations’. It isn’t clear what regulations Republicans consider ‘burdensome’. Many of the Biden administration’s measures are about information gathering to assess how the government can effectively use AI and evaluate what new standards and tools are necessary. A lack of information and standards would hinder the US government’s ability to deploy AI.

Repealing Biden’s executive order would likely have minimal impact on the private sector as only companies developing the largest models are impacted and the reporting requirements are built on cybersecurity measures and testing that companies are already doing.

Instead, its repeal would hinder efforts to promote civil rights, equity and non-discrimination in AI development. Republican policy would also ultimately complicate visa processes key to attracting foreign talent. This may serve another of Trump’s policy goals — decoupling with China. The Republican platform wants to secure strategic independence from China, possibly through massive tariffs. The ‘Project 2025’ initiative linked to the Trump administration wants to end economic engagement with China. This would harm both countries’ semiconductor industries, talent engagement and the overall relationship.

US companies selling less advanced chips and those doing final chip processing in China would be hit, as would Chinese companies making more basic ‘legacy chips’ for export, but the relative impact on each economy is unclear. Impeding Chinese AI talent moving to the United States would be a massive own goal, as Chinese-born researchers make up 38 per cent of top US AI researchers. Keeping China’s superior talent pool out of the United States would hobble US innovation while fuelling China’s development of AI.

Dialogue, not decoupling, is crucial to avoid stoking tensions that could lead to broader conflict. While Democratic policy has not been friendly to China, it has shown commitment to common-sense measures that benefit the United States and its relationship with China.

The future of AI development and US–China competition depends on the outcome of the November election. China has a long-term plan to secure AI dominance that — barring unexpected economic or political turmoil — it will continue to execute. The United States’ rhetoric and policy on AI have changed with the priorities of each new administration since the Obama presidency.

A Harris victory would see changes commensurate with a new administration but would likely continue to encourage responsible AI development while preserving civil rights. A Trump victory would likely promote full-throttle innovation but repealing Biden’s executive order would harm the government’s use of AI. Republican opposition to engagement with China would also hamstring the US semiconductor industry, international supply chains and the country’s talent pool while increasing the risk of escalated conflict that would cause irreparable harm far beyond AI development. While some vocal venture capitalists and Silicon Valley elites may laud a second Trump term, it would be a Pyrrhic victory that damages the very industry they claim to promote.

• About the author: Emmie Hine is a research associate at the Yale Digital Ethics Center and a PhD candidate in the Department of Legal Studies at the University of Bologna.
• Source: This article was published at East Asia Forum