Inside White House AI Policy

John is joined by Sriram Krishnan, former Senior White House Policy Advisor on Artificial Intelligence.  They discuss the Trump administration's AI development policies, including the administration’s AI Action Plan which Sriram largely developed. 

 The United States remains the global leader in artificial intelligence but its advantage over China is more narrow than many assume. Sriram believes that maintaining American leadership in the field requires reducing regulatory burdens, expanding access to computing resources, strengthening domestic infrastructure, and ensuring that allies have access to American technology rather than restricting its distribution. In short, the government should “let Silicon Valley cook."

While the administration has favored deregulation, it has also intervened to slow the release of some particularly advanced AI models. These positions are consistent because the government wishes to encourage rapid innovation but also protect critical infrastructure. Models with advanced cyber capabilities that could expose vulnerabilities in banking systems, power grids, military systems, healthcare networks, and other essential services will face heightened scrutiny.

At the same time, state regulation of AI is fragmented. The administration believes that the current patchwork of fifty different regulatory systems creates excessive compliance burdens and hampers innovation. That is why the administration is attempting to pass national AI legislation. The proposed legislation emphasizes four principal areas: intellectual property and the protection of creators' rights, the impact of data centers on local communities, protections for children, and the reduction of harmful bias and censorship.

John and Sriram also discuss the potential development of a voluntary pre-review process for new AI models that present particularly high cybersecurity risks. This process is being developed by the administration in close cooperation with the major AI companies. 

World leaders, including President Xi of China, have suggested that AI be regulated by an international body. Several leading figures in Silicon Valley have endorsed this idea. However, Sriram believes that it would be a mistake to allow an international bureaucracy which may not have American interests at heart control the cutting-edge technology currently being developed in the U.S. A different proposal put forward by Demis Hassabis would establish a Self-Regulatory Organization composed of the most significant AI developers. The proposed SRO would establish recognized industry standards.

Although the United States is the overall leader in artificial intelligence, Chinese companies are increasingly formidable competitors, particularly in open-weight models. The next several years will likely be critical in determining the global balance of power in artificial intelligence.

Podcast Link: Law-disrupted.fm
Host: John B. Quinn
Producer: Alexis Hyde
Music and Editing by: Alexander Rossi