ACUS, Stanford Law School, and NYU School of Law Announce Report on Artificial Intelligence in Federal Agencies

FOR IMMEDIATE RELEASE

Contact: Harry M. Seidman, Administrative Conference of the United States, (202) 480-2085, hseidman@acus.gov; Stephanie Ashe, Stanford Law School, (650) 723-2232, sashe@law.stanford.edu; Michael Orey, NYU School of Law, (914) 330-5555, michael.orey@nyu.edu

Washington, D.C., Stanford, Calif., and New York, February 18, 2020 — The Administrative Conference of the United States (ACUS), Stanford Law School, and New York University School of Law are pleased to announce the release of a major report exploring federal agencies’ use of artificial intelligence (AI) to carry out administrative law functions. This is the most comprehensive study of the subject ever conducted in the United States.

The report, entitled Government by Algorithm: Artificial Intelligence in Federal Administrative Agencies, examines the growing role that machine learning and other AI technologies are playing in federal agency adjudication, enforcement, and other regulatory activities. Based on a wide-ranging survey of federal agency activities and interviews with federal officials, the report maps current uses of AI technologies in federal agencies, highlights promising uses, and addresses challenges in assuring accountability, transparency, and non-discrimination.

Stanford Law School Professors David Freeman Engstrom and Daniel Ho, NYU Law Professor Catherine Sharkey, and California Supreme Court Justice Mariano-Florentino Cuéllar served as principal advisors on the report. They received research assistance from 30 Stanford law, computer science, and engineering students, and five NYU Law students, who participated in the Spring 2019 Stanford policy lab, Administering by Algorithm: Artificial Intelligence in the Regulatory State. Stanford’s Institute for Human-Centered Artificial Intelligence, with which Engstrom, Ho, and Cuéllar are affiliated, also provided seed funding for the report.

“We are at the dawn of a revolution in how government uses AI to do its work,” said Engstrom, a lead author on the report. “Our report — developed by talented law and engineering students in partnership with leading federal administrative agencies — offers a mix of broad insights and rich technical details that can guide government officials, judges, and Congress in ensuring these new technologies of governance serve the public good.”

“Drawing on Stanford's world-class schools of law and engineering, this report unearths the current broad and uncoordinated use of AI by our government,” said Ho. “By marrying our legal and technical expertise to forecast where technology is headed, we have produced a guiding report that can set the agenda for government in the age of AI."

“Whether they’re working to protect the environment or to limit illegal behavior in the marketplace, many federal officials understand the need for innovation to better serve the public –– and AI is a major part of that,” said Cuéllar. “But they’re also starting to grapple with difficult questions about how to define best practices and avoid risks involving AI. We hope this report will help agencies learn how to make the most of the possibilities and limit the risks.”

“Our probing exploration of AI inside the federal administrative state highlights the internal capacity-building efforts of agencies, which I think is the key to understanding how AI will shape the future of agencies’ monitoring health and safety risks and modern rulemaking efforts,” said Sharkey. “The study also features a unique exploration of how these changes may affect courts’ responses, when the inevitable challenges to rulemaking are brought before them.”

ACUS commissioned the report as part of its statutory mission to study the “efficiency, adequacy, and fairness” of the administrative process and publish reports that are “useful for evaluating and improving administrative procedure,” including “the use of science in the regulatory process.”

“ACUS is honored that so many distinguished scholars have chosen to collaborate with ACUS on this important initiative,” said Matthew L. Wiener, ACUS Acting Chairman and Vice Chairman. “ACUS thanks them, and the burgeoning scholars who participated in Stanford’s policy lab, for their many contributions to ACUS’s important work in improving the fairness and efficiency of federal administrative programs.”

Government by Algorithm is available here.