Musings on AI Policy from a Former U.S. Senate Staffer

In this talk, Simons Institute Law and Society Fellow Serena Booth will discuss the challenges of crafting specifications for AI systems; how easy it is for these systems to go rogue, whether through misspecification or other means; and her research studying both experts' and nonexperts' specifications. To this end, she will discuss her work in the U.S. Senate on AI policy. She will make the argument that consumer protection laws are an inalienable first defense for AI safety, and she will show how these laws have been successful in curtailing unsafe AI development and deployment to date.