Why “Beating China” in AI brings its own risks

The Biden administration introduced new export restrictions scheduled for this week Monitor the progress of AI globally and ultimately prevent the most advanced AI from falling into Chinese hands. The rule is the latest in a series of measures by Donald Trump and Joe Biden to rein in Chinese artificial intelligence.

With outstanding AI figures including OpenAIs Sam Altman and anthropic Dario Amodei The Trump administration, which has warned of the need to “beat China” in artificial intelligence, could make matters worse.

Paul Triolo is a partner at DGA Group, a global consulting firm, a member of the council on foreign affairs, and a senior advisor to the Penn Project on the Future of US-China Relations at the University of Pennsylvania. Alvin Graylin is an entrepreneur who previously managed China operations for a Taiwanese HPC electronics firm. Together, they monitored China’s artificial intelligence industry and the impact of US sanctions. In an email exchange, Triolo and Graylin discussed recent sanctions, Silicon Valley rhetoric, and the dangers of viewing global AI as a zero-sum game.

This interview has been edited for clarity and brevity.

What do you think again AI diffusion rule From the US government this week aiming to limit China’s access to AI?

Paul Triolo: In general, it focuses on high-performance computing clusters. The rule also controls specific model weights for the most advanced “frontier” models, but it is unclear how performance levels will be determined, and most open-weight (freely shared) AI models are tuned and improved by users, including major AI companies. in China.

Complex regulation and uncertain eligibility requirements bring considerable uncertainty to the long-term plans of both mid- and major US and western hyperscalers.

For hyperscalers such as Google, Microsoft, AWS and Oracle, the rule presents critical issues including slowed or more complex international expansion, new compliance and legal costs, impact on global R&D, and unclear enforcement requirements.

How have previous measures, including sanctions imposed by the first Trump administration, affected the AI ​​industry there?

Paul Triolo: US export controls have slowed China, but high-level sanctions have consolidated the will and efforts of the Chinese government. It has spent tens of billions to help domestic players capture technological or scale potential in key areas, resulting in significant changes to the semiconductor industry and its ability to support advanced hardware to develop frontier AI models.

Chinese AI developers have been very good at using older AI hardware from western firms and gradually integrating local alternatives into the development process. Chinese firms will continue to innovate in AI hardware and software, albeit not at the pace of their Western counterparts.

Why do you think so many people in Silicon Valley are talking about the need to “Beat China” in AI?

Paul Triolo: There is a growing relationship between conservative venture capitalists, mostly based in Silicon Valley, and tech companies whose business models depend on exposing the China threat. It’s a troubling combination of the China threat, self-interest, and pushback against regulation of advanced AI. It also describes Chinese competition around US AI as zero-sum, which is particularly dangerous.