Download this commentary (PDF)
In brief ...
A new legislative proposal aims to implement recommendations of the Administrative Conference of the United States to better handle mass campaigns and computer-generated inputs in the public commenting process.
Technology, including artificial intelligence, presents both opportunities and dangers for the public commenting process that is central to the making of regulatory policy. On the one hand, technology makes it easier for individuals and organizations to monitor agency rules and submit feedback. On the other hand, uses of technology such as mass comment campaigns and computer-generated comments run the risk of making it more difficult for agencies to identify unique information emanating from human beings interested in and affected by proposed rules. Technology, in other words, has the potential to overwhelm agencies with massive amounts of identical inputs as well as feedback created by algorithms.
Concerned about such negative possibilities, Congress has recently taken steps toward addressing the impacts of technology on public commenting. On March 5, the Comment Integrity and Management Act was introduced in the House of Representatives. Two days later, the bill—sponsored by Rep. Clay Higgins (R-LA)—was reported out of the Committee on Oversight and Accountability with bipartisan support.
The bill traces its origin to recommendations issued several years ago by the Administrative Conference of the United States (ACUS), “an independent federal agency in the executive branch charged with identifying and promoting improvements in the efficiency, adequacy, and fairness of the procedures by which federal agencies conduct regulatory programs.” These recommendations—titled “Managing Mass, Computer-Generated, and Falsely Attributed Comments”—in turn were derived from a report commissioned by ACUS that was written by a team of seven researchers with vast academic and practical experience at the intersection of technology and public commenting. With this pedigree, the bill is a welcome reflection of the merger of expertise in administrative law and political science with congressional concern about an important process through which Americans participate in government decision making.
The bill amends the E-Government Act of 2002 in four main respects. First, the bill requires agencies to determine, to the extent practicable, if comments are submitted by human beings (as opposed to being generated by computers) and if comments are part of mass comment campaigns (as opposed to being stand-alone submissions). Second, the bill allows agencies that receive mass comment campaigns to make one illustrative comment available in the rulemaking docket rather than post every single comment. Agencies that follow this practice must provide information about the size of mass comment campaigns (i.e., how many identical or nearly duplicate comments were submitted). Third, the bill requires the Office of Management and Budget (OMB) to develop guidance on how agencies should use technology tools to help make and publicize determinations about the occurrence of mass comment campaigns and computer-generated comments. Fourth, the bill requires the Government Accountability Office (GAO) to issue a report on ways to identify computer-generated comments, as well as their prevalence in and impacts on the rulemaking process.
A number of benefits follow from the bill’s prompting of agencies to identify mass comment campaigns and computer-generated comments. The posting of a single, illustrative mass comment—a practice that has been used for years by the Environmental Protection Agency—streamlines agency dockets without meaningfully reducing transparency about public participation. Although agencies are mandated under the law to respond to the substance of comments irrespective of the identity of the submitter, it is nevertheless the case that sorting computer-generated comments from those directly authored by human beings is a valuable practice. Such sorting will help make clear the extent to which, if at all, algorithms are capable of generating information that is unique and that meaningfully addresses legal, economic, scientific, and technical issues that agencies are expected to respond to in the process of developing rules.
The bill, to be sure, imposes search and reporting requirements upon agencies. The assistance of OMB and GAO (organizations with government-wide, procedural missions) in developing guidance and providing technical support, however, will mitigate these agency burdens. In the end, the bill identifies a danger posed to the public comment process by technologies such as artificial intelligence—the drowning out of unique human voices in a deluge of mass comment campaigns and computer-generated comments. Rather than throw the baby out with the bathwater, the bill wisely takes incremental steps that leave intact the underlying process of participation that has proven useful in regulatory policymaking ever since the 1946 passage of the Administrative Procedure Act.