On July 25, Stuart Russell testified on AI benefits, risks, and regulations at the U.S. Senate hearing titled “Oversight of A.I.: Principles for Regulation.”
Watch the hearing above or download his written testimony below.
Excerpt from the beginning of Stuart Russell's written testimony
Thank you, Chair Blumenthal, Ranking Member Hawley, and members of the Subcommittee, for the invitation to speak today. I am primarily an AI researcher, with over 40 years of experience in the field. I am motivated by the potential for AI to amplify the benefits of civilization for all of humanity. My research over the last decade has focused on the problem of control: how do we maintain power, forever, over entities that will eventually become more powerful than us? How do we ensure that AI systems are safe and beneficial for humans? These are not purely technological questions. In both the short term and the long term, regulation has a huge role to play in answering them. For this reason, I and many other AI researchers have greatly appreciated the Subcommittee’s serious commitment to addressing the regulatory issues of AI and the bipartisan way in which its work has been conducted.
Executive summary
-
- Artificial intelligence has a long history and draws on well-developed mathematical theories in several areas. It is not a single technology.
- Many current systems, including large language models, are opaque in the sense that their internal principles of operation are unknown, leading to severe problems for safety and regulation.
- Progress on AI capabilities is extremely rapid and many researchers feel that artificial general intelligence (AGI) is on the horizon, possibly exceeding human capabilities in every relevant dimension.
- The potential benefits of (safe) AGI are enormous; this is already creating massive investment flows, which are only likely to increase as the goal gets closer.
- Given our current lack of understanding of how to control AGI systems and to ensure with absolute certainty that they remain safe and beneficial to humans, achieving AGI would present potential catastrophic risks to humanity, up to and including human extinction.
- It is essential to create a regulatory framework capable of adapting to these increasing risks while responding to present harms. A number of measures are proposed, including basic safety requirements whose violation should result in removal from the market.
