Op-Ed (FedScoop): House ‘flash report’ on AI signals capacity gains

The following op-ed, authored by POPVOX Foundation’s Cofounder and Executive Director, Marci Harris, and its Director of Government Innovation, Aubrey Wilson, was published by FedScoop on September 18, 2023.

A new AI report from the Committee on House Administration underscores that Congress is not just looking at how to regulate AI but also working to ensure the responsible use of automated technologies within the legislative branch

Last week was an “AI-heavy” week on Capitol Hill — with an AI Forum hosted by Senate Majority Leader Chuck Schumer, D-N.Y., a number of AI-related hearings in the House and Senate, and a Congressional Hackathon that kicked off with AI-focused remarks from both Speaker Kevin McCarthy, R-Calif., and former House Majority Leader Steny Hoyer, D-Md.

With all that, one AI announcement on Thursday may have escaped notice in the hoopla, but it is a tangible step demonstrating that Congress is not just looking at how to regulate AI but also working to ensure the responsible use of automated technologies within the legislative branch.

On Thursday, the Committee on House Administration (CHA) released a flash report through its new Modernization Subcommittee on the use of AI within the House of Representatives. As the committee tasked to oversee the day-to-day operational and institutional offices that support Members and their staff, CHA has jurisdiction over House technology. And with the addition of a new bipartisan Modernization Subcommittee at the beginning of the 118th Congress, it is not wasting any time. 

With the advent of ChatGPT and other large language models (LLMs), CHA recognized that the House would be faced with big questions about appropriate use and potential risks and opportunities. Leveraging existing talent and expertise from within the legislative branch, CHA brought on a nonpartisan detailee from the Government Accountability Office’s Office of Science, Technology Assessment, and Analytics Team (STAA) to aid policymakers in understanding the new technology.

GAO details were common in previous decades but are increasingly rare in the modern Congress. However, bringing in-house nonpartisan subject expertise is a notable capacity builder, resulting in a more agile Congress that can get smarter, faster on emerging issues. 

As noted in the Flash Report: CHA has asked the leaders of congressional support entities to begin thinking about transparent AI use case inventories and comprehensive AI-related governance documents.

The flash report is noteworthy in several ways:

• Congress is addressing its internal “pacing problem”: While the headlines are focused on policy questions and potential regulation, CHA is working to ensure that AI adoption proceeds at a safe, but active pace. 

• Congress is also addressing its inter-branch “pacing problem,” in which Congress fails to keep up with the executive branch and the agencies of government it is tasked with overseeing: AI has the potential to level the playing field between the branches. AI-boosted workflows could dramatically improve the internal operations of the U.S. House, increase the service levels of the legislative branch agencies, drastically improve the support members are able to provide to their constituents, and help Congress better process data for insights into federal programs.

• Congress is already experimenting with AI:

  • The “Comparative Print Suite” uses Natural Language Processing to help legislative staff compare what’s changed between different bill drafts

  • The CAO has established an AI Working Group to gather feedback from a limited number of ChatGPT license holders in Member offices.

  • The Library of Congress, through LC Labs, is researching how machine learning can support libraries.

• Congress isn’t starting from scratch, nor is it building its AI framework in a vacuum. Instead of reinventing the wheel, CHA is learning from strategies, standards, and guidelines established in other areas of government — including NIST’s AI Risk Management Framework, the bipartisan AI in Government Act of 2020, Executive Order 13960: “Promoting the Use of Trustworthy AI in the Federal Government,” and resources from the Office of Management and Budget, the National Artificial Intelligence Advisory Committee (NAIAC), and the Government Accountability Office. 

• The House is using AI as a momentum-builder for improved data hygiene — including standardization, maintenance, and transparency. Changes that civil society has called for over many years are getting a fresh look thanks to the new opportunities presented by AI. Just as the pandemic jump-started digitization leaps like the ability to introduce bills digitally through the eHopper or to gather digital signatures through Quill, so new AI tools are catalyzing new discussions about the importance of data standardization and data management that will ultimately lead to better access to institutional resources for congressional Members, staff and the public.

This is what “modern” looks like.

The House began its most recent “modernization” journey with the creation of the bipartisan House Select Committee on Modernization (ModCom) in January 2019. The select committee was reauthorized in 2021 and over the 116th and 117th Congresses, members produced 202 bipartisan recommendations, many of which are well on their way to implementation. 

But one theme that emerged throughout the work of the ModCom was that “modernization” was not a destination to be reached but rather an ongoing practice of continuous improvement. In fact, of the 202 recommendations produced in December of 2022, there is only one passing mention of artificial intelligence. But the House is not simply implementing recommendations. Through the new Modernization Subcommittee in CHA, the work continues and members and staff continue to learn, to respond to new circumstances, and to anticipate an ever-changing future.