A critique of the call for public AI
As I spend more time in the AI infrastructure business, I've been thinking more and more about the government's role in AI . There's no shortage of opinions and position papers on the topic, and sadly, most of them are written from the perspective of the government rather than the AI industry. As a result, they are often full of misunderstandings, misleading statements, or ideas about the world that are six months out of date—the blink of an eye to the government, but an eternity in the AI industry. The latest to cross my desk is a report titled The National Security Case for Public AI which was released on September 27, 2024 by the Vanderbilt Policy Accelerator. At a high level, the authors try to make the case that the U.S. government should build its own vertically integrated AI stack (from silicon to data centers to applications) to compete with (“complement”) effort in private industry. They also suggest regulating the AI industry in a way analogous to how public
