The CHIPS and Science Act, along with the Biden administration’s export controls for semiconductors, have set the stage for a comprehensive industrial policy on advanced semiconductors in the United States. Now, a similar discussion is beginning to unfold around artificial intelligence (AI). While still in its early stages, federal officials are looking at a promote and protect approach to AI, focusing on national security concerns.
Government funding for domestic nondefense AI innovation would be a key driver in promoting the development of AI technologies. The Senate’s bipartisan working group has called for appropriations of at least $32 billion per year for non-defense AI innovation, a number recommended by the National Security Commission on Artificial Intelligence. However, the proposal has yet to be turned into a bill, and its passage through Congress is uncertain, especially given the high cost involved.
With few significant bills expected to pass Congress in the near future, the timeline for AI funding provisions to gain traction is limited. The Senate working group seems more focused on building a genuine bipartisan consensus rather than rushing through partisan legislation. The House also has its own bipartisan AI task force, which will provide additional perspective on addressing the new technology.
The other aspect of the industrial policy revolves around protecting advanced AI models from foreign adversaries, particularly China. The Commerce Department is considering export controls similar to those imposed on semiconductors, aiming to limit access to the backend software powering the models. The logistics of enforcing these controls pose a challenge due to the intangible nature of the technology involved.
Determining which AI models would fall under export restrictions is still a critical aspect that needs to be clarified. Potential standards could include computing power requirements or considerations about the type of data used and the model’s intended use. While these efforts are in their early stages, the focus on AI among officials suggests that this industrial policy could take shape over the next few years, similar to the path taken by the CHIPS and Science Act.
In conclusion, the discussions around government funding and export controls for AI are essential in shaping the regulatory landscape for the technology in the coming years. These discussions will not only determine lawmakers’ priorities for AI but also highlight any partisan divisions that could hinder legislative progress. As with the semiconductor policy, the development of an industrial policy for AI is likely to be a prolonged process that will require continued attention and engagement from stakeholders.
