Human-AI Collaboration workshop @ ACM EC · 2025
Designing AI for Meta-Expertise in Regulatory Drafting
This paper asks how AI writing assistants can assist federal rule writers. Draft rules must ground authority in statute, reflect current science, and persuade diverse audiences. Drafting such rules demands what Frank Pasquale calls meta-expertise: the ability to arbitrate among competing expert domains, in this case, between legal precision, scientific accuracy, and persuasiveness. I argue that designing for this meta-expertise is key to building systems that support rule writers’ expert judgment.