A Better Approach to AI Regulation: Why a Sovereign Wealth Fund Isn't the Answer

The Problem with AI Regulation

Bernie Sanders has proposed creating a US sovereign wealth fund to regulate AI companies. While the goals of this proposal are laudable, the authors argue that there are better ways to achieve them.

The Risks of Public Ownership

The authors argue that public ownership of AI companies could lead to the government prioritizing corporate profits over public interests. They cite the example of Norway's sovereign wealth fund, which has not been effective in steering oil companies towards pro-environmental policies.

A Better Approach: Taxation and Public AI

The authors propose alternative solutions, including taxation and a public AI option. They suggest that an excise tax on datacenters' energy use or an AI token tax could be effective in sharing private rewards with the broader society. They also propose a public AI option, where governments establish publicly developed and operated AI models run by public institutions under democratic control.

The Swiss Model: A Public AI Option

The authors cite the example of Switzerland's Apertus project, a large language model built by Swiss public servants and researchers. They argue that this approach can provide a competitive baseline for private AI offerings and encourage responsible behavior.

Conclusion

The authors urge Sanders and other political leaders to consider alternative approaches to AI regulation, including taxation and a public AI option. They argue that these approaches have a greater chance of influencing corporate behavior towards the public interest.