The Biden administration has introduced new export restrictions aimed at controlling AI’s global progress and preventing advanced AI from falling into China’s hands. The rule is part of a series of measures put in place by previous administrations to keep Chinese AI in check.
Prominent AI figures, including OpenAI’s Sam Altman and Anthropic’s Dario Amodei, have warned of the need to “beat China” in AI. However, experts argue that this zero-sum approach is problematic and may escalate tensions further.
Paul Triolo, a partner at DGA Group, believes the new rule focuses on high-performance computing clusters and proprietary model weights but has unclear compliance conditions. The complex regulation injects uncertainty into US and Western hyperscalers’ long-term plans.
US export controls have slowed China’s AI development but have also unified the Chinese government’s efforts to become more self-reliant. Chinese developers have adapted by leveraging legacy AI hardware from Western firms, and domestic alternatives are becoming increasingly sophisticated.
The “beat China” narrative in Silicon Valley has raised concerns about conflating personal gain with national security interests. Elon Musk, who has called for international cooperation on AI governance, may push back against this approach.
Experts Alvin Graylin and Paul Triolo caution that the US cannot maintain a sustainable lead over China in AI development. Collaborative research has been crucial to progress, but the dominant paradigm driving US policy is based on assumptions of future conflict.
The duo advises the incoming president to pivot towards collaboration, establishing common AI governance standards, monitoring misuse globally, and supporting academic-industry collaborations across borders. A global effort akin to CERN for AI could bring more value than a Manhattan Project for AI.
Source: https://www.wired.com/story/why-beating-china-in-ai-brings-its-own-risks