Download this commentary (PDF)
The George Washington University Regulatory Studies Center and Norm Ai hosted a conference exploring the question, “Can AI Streamline Regulation and Reduce Compliance Burdens?” Speakers from across the political spectrum—representing government, academia, the private sector, and nonprofit sector—discussed appropriate roles for AI in government, as well as principles to consider when assessing approaches to AI regulation.
Several themes emerged. First, the term artificial intelligence or “AI” encompasses a range of activities, from simple text analysis and machine learning to sophisticated agentic models. The use cases are varied, as well. Government regulators are using AI in their supervisory roles (“sup tech”); states are using it to identify existing regulations that may be inconsistent with legal authorities or more burdensome than other states; and small businesses are finding that AI can assist them in meeting licensing and other regulatory responsibilities, which can be uniquely challenging for small businesses that do not have large legal and compliance departments.
Second, while AI can have comparative advantages in efficiency, speed, responsiveness, consistency, and predictability, human involvement in governmental uses of AI will remain important for accountability and legitimacy. For example, AI agents can provide a valuable screening function or offer first line compliance assistance, but for writing binding regulations or bringing enforcement actions, the human element remains essential.
Although speakers had different views as to the potential value and risk of evolving AI tools and applications, those did not break down along party lines. For example, Democratic Commodity Futures Trade Commissioner Kristin Johnson spoke to the potential of AI to enhance regulators’ supervisory effectiveness and ability to detect fraud.
With respect to regulating private sector use of AI, several speakers recommended risk-based approaches that allowed the emerging technology to innovate and evolve. For the most part, regulators should be able to rely on existing authorities that are technology neutral to address potential risks. To the extent that some novel use cases pose unique harms, new legislation or policies may be needed.
Overall, panelists agreed that AI offers vast untapped potential for government efficiency while highlighting the need for managing risks and maintaining public trust.