Congress is as prone to fads as the rest of humanity. For example, today there will be two hearings on artificial intelligence (AI) in the Senate. The Committee on Homeland Security and Governmental Affairs will hold one on Artificial Intelligence in Government. (I resisted every one of the possible corny jokes about that one; please reciprocate.) That should be relatively tame, but the Judiciary Committee’s Subcommittee on Privacy, Technology, and the Law will have Oversight of A.I.: Rules for Artificial Intelligence. This will feature Samuel Altman, the CEO of OpenAI, the developer of the ChatGPT chatbot that has generated so much attention. Finally, on Wednesday the House Judiciary Committee will hold a hearing on Artificial Intelligence and Intellectual Property: Part I – Interoperability of AI and Copyright Law.

These hearings are indicative of a broader phenomenon: D.C. is obsessed with AI, both in Congress and in the agencies, leading to pressure to legislate, regulate, or otherwise dictate the development of AI. This rush to regulate is problematic, as there is a lot to be learned about AI and premature regulation risks forgoing large benefits.

To begin, not all AI is the same. ChatGPT, for example, is an example of generative AI, which can create new content based on existing data. In contrast, AAF recently met with a long-standing supplier of predictive AI, which analyzes data to make predictions about future events. The particular example was using AI to match callers with service representatives to maximize the chances of a good service experience. At present, a real risk is to treat all AI as the same thing and subject it to a single, ill-fitting regulatory regime.

The second point is that there already exists a lot of capability to regulate AI. (Adam Thierer makes this point nicely.) That’s because AI is typically an input to some other goal (e.g., customer service) and there is broad regulation of an enormous range of goods, services, and activities. As long as there is no problem with the outcomes, there is no particular reason to regulate the inputs.

Expect a lot of fireworks in today’s hearings and beyond. Expect a lot of broad, sweeping claims about the likely future of AI. But don’t expect much of it to be well-informed or accurate. Finally, expect a lot of talk about legislation and regulation, but hope that very little of it comes to fruition anytime soon.