Congress Scrambles to Steer AI Development
While talk of artificial intelligence has been humming in Congress for years, this spring’s launch of powerful language-generation models and chatbots has triggered a frenzy of activity on Capitol Hill.
In June, Senate Majority Leader Chuck Schumer (D-NY) announced that he plans to develop legislation to steer the development and use of AI, adhering to what he calls a “SAFE Innovation Framework.” The goal: Develop AI technology in ways that safeguard national security, ensure accountability for misuse, align AI systems with democratic values, and ensure AI systems can explain how they draw conclusions.
Schumer plans to convene “insight forums” starting this September to supplement more traditional committee hearings. “We need the best of the best sitting at the table — the top AI developers, executives, scientists, advocates, community leaders, workers, national security experts — all together in one room, doing years of work in a matter of months,” he said. Schumer has picked Sens. Martin Heinrich (D-NM), Mike Rounds (R-SD), and Todd Young (R-IN) to help lead the Senate’s development of AI legislation. Heinrich and Rounds co-chair the Senate AI Caucus, and Young was the lead Republican co-sponsor of Schumer’s innovation policy push last Congress, which resulted in the CHIPS and Science Act.
Pieces of potential bills have begun to emerge. In July, Heinrich and Rounds introduced legislation to create a National AI Research Resource (NAIRR), which would give US researchers access to data and computational resources necessary to develop new AI technologies. The legislation envisions the National Science Foundation as the lead agency for the NAIRR, following the recommendations of a congressionally mandated task force. Senators have also attached a host of AI-focused proposals to their versions of the annual National Defense Authorization Act, many of which are likely to make it into the final version of the bill negotiated with the House.
House Speaker Kevin McCarthy (R-CA) has not announced an analogous AI initiative, but several House committees have taken up the subject. In June, the House Science Committee held the first of a series of hearings on AI to inform a potential follow-on to the National AI Initiative Act of 2020, which the committee helped develop. Committee member Jay Obernolte (R-CA), who has a master’s degree in AI, is sponsoring the House version of the NAIRR legislation.
Meanwhile, science agencies have been pitching ideas to Congress on how they could contribute to a national push on AI. In particular, the Department of Energy published a report this summer that calls for DOE to tap its deep expertise in supercomputing to develop AI tools focused on scientific discovery, energy technology development, and national security. One of the lead authors, Rick Stevens of Argonne National Lab, participated in a Senate briefing on AI in July, along with DOE Under Secretary for Science and Innovation Geri Richmond and NSF Director Sethuraman Panchanathan.
Stevens told a DOE advisory committee this summer that the initiative is designed as a follow-on to DOE’s roughly $2 billion program to deploy exascale computers, which will wrap up soon. The AI initiative could be much larger than the exascale effort, ultimately involving thousands of national lab scientists, Stevens said.
Asked how the initiative would be funded, given recent budget caps set by Congress, Stevens noted that Congress could provide emergency funding. Schumer has stated he will pursue follow-on legislation to the CHIPS and Science Act, which included $52 billion dollars for the semiconductor sector.
Mitch Ambrose is the director of science policy news at the American Institute of Physics.