Subcommittee on Modernization and Innovation:
“The Future of Constituent Engagement with Congress”
Chairwoman Bice, Ranking Member Torres, and Members of the Subcommittee, thank you for the opportunity to testify today.
Full testimony of Beth Simone Noveck
At precisely the moment when Congress is most overworked, addressing complex and interdependent challenges—when committee staff are roughly 40 percent smaller than they were in 1980, when CRS, GAO, and CBO have shrunk by 45 percent since the 1970s—public engagement can sound like an unreasonable burden to congressional staff already stretched thin.
After all, it is hard enough to field the more than 81 million calls Congress receives from constituents every year.
Historically, meaningful participation has been too expensive, too time-consuming, and too often lacked the structures needed to make public contributions usable for decision-makers. On the web, the more people talk, the harder it can be for institutions to listen.
But as we have learned in the State of New Jersey, pairing the right artificial intelligence tools with disciplined process design makes it possible for government institutions to listen better, learn faster, and govern more effectively—even at scale.
Two recommendations from this Subcommittee provide a practical starting point for that transformation.
Recommendation 176 calls for exploring ways to solicit public input for committee hearings. Long before generative AI, Brazil’s Federal Senate began inviting residents to submit questions for committee hearings by phone and through the web.
Today, the Senate has begun integrating AI tools that help staff de-duplicate submissions, cluster similar questions, and highlight those that directly address a committee’s oversight goals. Participation has been substantial—46,000 questions across 546 hearings in 2023 and 69,000 questions across 440 hearings in 2024. Public input has become so routine that, as the head of the e-Citizenship office notes, “If we are late, committee chairs ask, ‘Where’s the list of questions?’”
This Subcommittee could begin testing Recommendation 176 through a simple, low-risk pilot conducted over the course of a few hearings that invites the public to submit questions.
Until recently, processing free-text submissions at scale was impractical.
Today, AI makes it straightforward to gather feedback by voice or text, remove duplicates, filter off-topic content, cluster questions into themes, and synthesize learnings.
I outline this process in my written remarks, along with additional opportunities for “crowdlaw,” or public engagement in lawmaking.
The AI for Impact students I teach—who work full-time embedded with government partners including the State of New Jersey and the City of Boston—recently built a free, open-source tool called Open Feedback, now being deployed on Boston.gov.
Residents submit feedback in natural language, and the AI assistant responds with clarifying questions to improve the submission. The tool allows staff to organize comments by topic, analyze patterns, and route issues to the appropriate department for faster feedback.
The next AI for Impact cohort begins in January, and we would be happy to adapt this work to support a congressional pilot.
Through InnovateUS—the peer-to-peer learning program we started in New Jersey, where public servants learn how to use AI to serve the public—we are learning from jurisdictions ranging from St. Louis and Bowling Green to Denmark and Germany about how they are using AI to listen at scale. Based on more than a dozen workshops focused specifically on AI and public engagement, we are distilling these lessons into a one-hour training we would be happy to share with congressional staff this spring.
Of course, Congress must guard against hallucinations by ensuring that AI is used for sorting and organization rather than decision-making.
The greater risk Congress must guard against is ineffective or performative engagement. Public participation disconnected from real decisions wastes time and erodes trust.
Over the long term, Congress can develop its own tools for engagement and legislative process management by using legislative data to build a publicly governed Congress.ai model.
But these pilots offer a practical, immediate way to tap the expertise and experience of the American public to strengthen lawmaking and oversight.
Watch the full Q&A of the session: “The Future of Constituent Engagement with Congress”