- Congressional Research Service exploring generative AI models
- Evaluation of models looks at accuracy, objectivity, performance
The Congressional Research Service is developing five generative AI models to help it summarize bills introduced in Congress. The hope is that eventually they can be used to write bill summaries on Congress.gov.
“We think there is potential,” Robert Newlen, the interim director of the CRS, said at a House Administration modernization subcommittee hearing Wednesday. “Our hope is we can implement this soon so that our staff can spend more time on the analysis, which is the really hard work, and less time on the summaries.”
The models could help clear a huge backlog of bill summaries, he said.
More than 17,800 bills were introduced in Congress during the 117th Congress, a two-year period from Jan. 2021 to Jan. 2023, according to govtrack.us. More than 13,700 bills have been filed so far in the current Congress.
“We’re very excited about artificial intelligence,” Newlen said. “But like everyone, we’re moving cautiously to ensure that any information that we get from AI meets all of our standards of nonpartisanship, authoritativeness.”
The move by CRS, which is part of the Library of Congress, comes even as Congress and the Biden administration grapple with the potential and pitfalls of generative AI.
In written testimony submitted to the committee, Newlen said that CRS is exploring products to help draft, review, and publish bill summaries, which are available to both Congress and the public. The evaluations looks at accuracy, objectivity, and performance, he said.
To contact the reporter on this story:
To contact the editors responsible for this story:
Learn more about Bloomberg Law or Log In to keep reading:
See Breaking News in Context
Bloomberg Law provides trusted coverage of current events enhanced with legal analysis.
Already a subscriber?
Log in to keep reading or access research tools and resources.