The bipartisan leaders of the House Administration Committee have started a dialogue on the chamber’s uses of artificial intelligence that tracks with risk management discussions across government and the private sector, with an emphasis on the need to implement effective governance structures such as the National Institute of Standards and Technology’s AI framework.
“Today’s hearing serves as the first of hopefully many public discussions about Artificial Intelligence in the Legislative Branch, and how this evolving technology could impact the people’s House and the work we carry out on behalf of our constituents,” House Administration Chairman Bryan Steil (R-WI) and ranking member Joe Morelle (D-NY) said in a joint statement prior to the committee’s Jan. 30 hearing.
“We are actively working in a bipartisan manner to understand the complexities of AI and how to implement guidelines that will support institutional improvements while ensuring continued safety, transparency, and accountability to the American people,” the committee leaders said. “We look forward to todays’ conversations and future discussions.”
The committee heard from Taka Ariga of the Government Accountability Office; John Clocker, deputy chief administrative officer for the House of Representatives; Judith Conklin of the Library of Congress; and Hugh Halpern, director of the Government Publishing Office. The witnesses all pointed to the NIST AI risk management framework as a key to proper AI governance.
"Congress must … ensure our legislative branch agencies are developing comprehensive AI governance plans. These plans are foundational IT best practices and are necessary to effectively manage AI over the long-term,” Steil said in his opening statement. "The NIST AI Framework is a critical resource that can help legislative branch agencies turn AI principles into specific governance policies."
Ranking member Morelle’s opening statement also highlighted the January 2023 NIST document.
“I’m grateful that the Committee and our institutional partners have utilized the Administration’s National Institute of Standards and Technology AI framework, and related executive orders, as models for the legislative branch as it relates to AI governance policies,” Morelle said. “With these as a guide, Congress I think will be better equipped to minimize risk and adopt this technology responsibly and ethically.”
Clocker, in response to questions from the panel, said the CAO’s office is developing “AI-specific” guidelines and protocols for providers of AI services to the House, but that more resources will be needed to evaluate the products. The alternative, he said, is to “limit the number of products we authorize.”
He said the office’s approach is based on the NIST AI framework, “tailored for our environment,” and that ultimately all House offices will use it. “We’ll have a draft policy to the [Administration] committee in two or three months,” Clocker said.
Rep. Terri Sewell (D-AL) in her questioning probed how the various offices are addressing bias and cybersecurity risks. Conklin said the Library of Congress addresses bias in its strategic plan on AI systems and “adheres to the NIST AI framework to evaluate biases.”
Clocker said the CAO has “very strong technology controls around our administrative data” and is proceeding cautiously on implementing AI tools.
Rep. Morgan Griffith (R-VA) sought information on AI tools that could help lawmakers do their jobs more effectively, such as faster transcription and bill summaries.
“This is going to be an ongoing conversation” about leveraging benefits and managing risks, Steil said in wrapping up the session. He added a note of optimism for the lawmakers, congressional staff and others who crowded the hearing room: “One thing I heard repeatedly is AI will not replace humans.”