Honest Brokers in AI Regulation: How Public Comments Introduce Evidence to Regulators
Public comments are written submissions from people and organizations responding to a proposed government rule. They can offer evidence, raise concerns, or recommend changes. In US agency rulemaking, this public input is one way people outside government can contribute to decisions. The article studies these comments in the context of potential AI regulation in the USA. For example, a commenter might cite research or refer informally to evidence relevant to a proposed AI rule. The researchers collected comments on 40 potential regulations and processed their texts to identify formal citations and informal references. They then examined the evidence commenters supplied and compared patterns across stakeholder groups. The article finds that comments do not all provide the same kind or amount of evidence. Business commenters supplied less evidence than other groups when comment length was held constant, while government, university, and civil society commenters provided significantly more research and legal sources. Comments therefore can make different sources of information accessible to regulators. The article does not detail each procedural step for submitting comments or how agencies weigh them.
What are public comments in the US rulemaking process?
Public comments are written submissions from people and organizations responding to a proposed government rule. They can offer evidence, raise concerns, or recommend changes. In US agency rulemaking, this public input is one way people outside government can contribute to decisions. The article studies these comments in the context of potential AI regulation in the USA.
For example, a commenter might cite research or refer informally to evidence relevant to a proposed AI rule. The researchers collected comments on 40 potential regulations and processed their texts to identify formal citations and informal references. They then examined the evidence commenters supplied and compared patterns across stakeholder groups.
The article finds that comments do not all provide the same kind or amount of evidence. Business commenters supplied less evidence than other groups when comment length was held constant, while government, university, and civil society commenters provided significantly more research and legal sources. Comments therefore can make different sources of information accessible to regulators. The article does not detail each procedural step for submitting comments or how agencies weigh them.
What did the researchers find about how much and what kinds of evidence different groups supplied?
The researchers found differences in both the amount and type of evidence supplied by stakeholder groups. They compared evidence while holding the length of comments constant, so the differences were not simply because some groups wrote longer submissions. Business commenters introduced less evidence overall.
Government, university, and civil society commenters supplied significantly more research and legal sources. The team identified evidence through formal citations and informal references in public comments. The publication status of cited research was nearly evenly split between preprints and peer-reviewed articles, and commenters from all types of groups cited both. Self-citation was also prevalent across groups, but a small cluster of organizations accounted for over half of all self-citations in the collection.
These results complicate the idea that public comments are simply a channel for information capture. They show that different commenters contribute different evidence, and that evidence comes in varied forms. The findings do not establish how regulators ultimately used particular citations or whether a given comment changed a rule. They do show that who comments can affect what information is presented to regulators.
How large was the study’s collection of proposed regulations and public comments?
The study assembled a corpus covering 40 potential regulations and 2,664 public comments. These comments were submitted by a variety of stakeholders in the USA. That collection gave the researchers a basis for comparing the evidence offered in public input on possible AI regulation.
The team computationally processed the texts to identify evidence in two forms: formal citations and informal references. They then compared how much evidence different groups supplied and what kinds of sources appeared. The analysis included research and legal sources, publication status of cited research, and self-citation patterns. Comment length was held constant when comparing the amount of evidence supplied by different groups.
The study found distinct patterns across commenters. Businesses introduced less evidence overall, while government, university, and civil society commenters provided significantly more research and legal sources. Preprints and peer-reviewed articles were nearly evenly represented among cited research. Self-citation occurred across groups, though a small cluster of organizations accounted for over half of all self-citations. The article does not provide further breakdowns of the corpus by regulation or stakeholder count.
What is the difference between an “Issue Advocate” and an “Honest Broker” in policymaking?
An Issue Advocate generally argues for a particular position or outcome in a policy debate. An Honest Broker, in common policy language, aims to clarify the available evidence and options without pushing one preferred answer. These descriptions help explain different ways commenters may participate in policymaking. The article uses these roles to interpret patterns in the evidence supplied through comments.
For example, an Issue Advocate might select evidence that supports a requested policy direction. An Honest Broker might seek to make a wider range of relevant evidence available to decision-makers. The article itself analyzes formal citations and informal references, including research and legal sources, but it does not classify each commenter as one role or the other. Nor does its abstract provide a detailed definition of these terms.
The distinction is therefore a useful lens, not a reported tally of two fixed groups in this study. Researchers found that commenters supplied different amounts and types of evidence, and that self-citation appeared across groups. Their findings complicate the idea that public comments alone are a means of information capture. The article suggests that commenter roles matter for how evidence enters the policymaking process, without claiming that evidence is automatically neutral or decisive.
How can evidence in public comments affect the information regulators use when shaping AI rules?
Evidence in public comments can give regulators information to consider when developing AI rules. The article focuses on how commenters make evidence accessible through formal citations and informal references. This matters because AI regulation is susceptible to capture, and the article identifies industry actors’ information advantage as part of that concern.
For example, the researchers found that government, university, and civil society commenters provided significantly more research and legal sources than business commenters, holding comment length constant. Cited research included both preprints and peer-reviewed articles, in nearly equal shares. By submitting sources and references, commenters can introduce material that may inform regulators’ understanding of potential rules and their effects.
The study measures evidence supplied, not the influence of particular comments on final decisions. Its findings therefore do not show that a citation changed a regulation or that regulators accepted every source. They do show that the information available through public comments depends partly on who submits them and what evidence they provide. The authors argue that these patterns illuminate an often overlooked part of policymaking and how information reaches AI regulators.
What is regulatory capture, and how can an industry’s information advantage contribute to it?
Regulatory capture is a general term for a situation in which a regulated industry gains undue influence over the rules meant to govern it. The article says AI regulation is susceptible to various forms of capture by industry actors who hold undue influence over policymaking. This definition of capture is established policy knowledge; the abstract does not give a formal definition.
An information advantage can contribute when companies developing a technology know more about how it works, its risks, or its likely effects than policymakers do. Regulators may then depend heavily on information supplied by industry, potentially giving industry voices greater weight. The article identifies this information advantage in AI development as one reason regulation may be vulnerable to capture.
However, the study’s findings complicate a simple story in which public comments themselves serve as a channel for information capture. Holding comment length constant, business commenters introduced less evidence overall, while government, university, and civil society commenters supplied significantly more research and legal sources. The study examines evidence in comments; it does not measure undue influence or prove that capture occurred.
How does US agency rulemaking work, from a proposed rule to a final regulation?
In general US agency rulemaking, an agency develops and publishes a proposed rule, then invites the public to comment. People and organizations can submit views, evidence, and recommendations. The agency reviews the feedback as it develops a final rule. These are general features of US rulemaking; the article’s abstract does not describe the procedural steps in detail.
For example, a proposed AI regulation could receive comments containing research citations, legal sources, or informal references. The researchers studied this kind of public input across 40 potential regulations. They computationally processed 2,664 comments to identify evidence intended to inform regulators. This illustrates how the comment stage can bring outside information into the rulemaking process.
The article focuses on the evidence people submit, rather than explaining the legal requirements for issuing a final rule or how agencies evaluate every comment. Its findings show that evidence varies by commenter type: business commenters supplied less overall, while government, university, and civil society commenters provided significantly more research and legal sources. The process provides a route for evidence to reach regulators, but the study does not establish how specific evidence shaped final regulations.
This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.
Read more in the JupiteX app
Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.
Or read more news on the web