Honest brokers in AI regulation: how public comments introduce evidence to regulators
Honest Brokers in AI Regulation: How Public Comments Introduce Evidence to Regulators
The researchers built a corpus of 40 proposed US AI regulatory rules and 2664 public comments, using computational methods to extract formal citations and informal references through which comments provide evidence to regulators. The study found significant differences across parties in the types and amounts of evidence submitted; controlling for comment length, business commenters overall submitted less evidence, while government, university and civil society commenters provided significantly more research and legal sources. The publication status of cited research was close to an even split, with preprints and peer-reviewed papers cited by all types of commenters. Self-citation was prevalent across all groups, but a small cluster of organizations contributed more than half of the self-citations in the corpus. The authors argue these results complicate the narrative that public comments alone constitute a means of information capture, and explain commenters' roles as either issue advocates or honest brokers.