Crowdsourcing + AI rulemaking

Crowdsourcing + AI is a viable strategy for rulemaking in an era of smaller agency staffs. This seems to be emerging thinking in the Administration.

The first sign was the deregulation suggestion form used with the spring's omnibus Deregulation RFI. The latest is last week's advance notice of proposed rulemaking (ANPRM) in the personal financial data (1033) reconsideration.

In the 1033 litigation, the CFPB promised the court an "expedited" reconsideration rulemaking. But how is using an ANPRM consistent with an expedited rulemaking?

The optional ANPRM phase is usually skipped even in non-expedited rulemakings. Agencies use it when they need to gather more information to write a detailed proposal. It's an odd choice when an agency is reconsidering a rule that the agency just issued and litigated.

But there is one way to reconcile use of an ANPRM with desire to expedite the rulemaking: The comments on an ANPRM supply the raw materials you need for an AI tool to assemble a credible proposed rule draft.

If you need to draft a proposed rule, but lack enough experienced personnel to handle all your rulewriting, an ANPRM followed by AI drafting is probably the fastest way to do it. You still need the experts, but you can go faster with fewer.

From experience, I believe this approach is technically feasible with current technology. Earlier this year, we created and used many of the building blocks you would need for such a solution to help a client respond to the Deregulation RFI. (Though we didn't go as far as drafting full proposals.)

The question is whether the Administration will bring together the right combination of domain experts and technical talent needed to write rules robust enough to survive the inevitable legal challenge. Frankly, their track record on that front has been mixed.