AI and Emerging Technology

As we embark upon a new era in artificial intelligence, we’re helping Congress stay ahead of the curve so it can properly regulate emerging technology platforms and further modernize its own systems to better serve the public.

We’re on the forefront of promoting the strategic early adoption of AI in the congressional workplace. These technologies are already having an impact on society and prompting debates about its merits and threats.

Through Hill briefings, our First Branch Forecast newsletter, and outreach to congressional offices, we’re pragmatically helping Congress be up to the task of understanding AI and responsibly overseeing the corporations rolling out these new technologies.

We’re equally excited about the potential benefits for improving productivity for the enormous workflow in the Legislative branch. AI’s most likely impact in Congress will be to alleviate some of the drudgery of managing incoming and outgoing communications, summarizing information, and mining congressional data for strategic insights. 

AI’s Large Language Models or “LLMs” can write useful first drafts of “Dear Colleague” letters, constituent communications, social media posts, arguments for or against a policy, and they can also write computer code, whether for a simple spreadsheet formula or in various computer languages.

This frees congressional staff to focus on more substantive work. Instead of drafting, they’re reviewing. Instead of data entry, they can focus on more human engagement and policymaking. It can also expand in-house capacity to create graphics and videos for hearings or social media engagement.

AI’s ability to quickly generate structured data and summarize it with little training will be game changing.

• For example, AI can extract information, such as a constituent’s name, or turn unstructured address information into structured information, which can then be used to more efficiently write and address personalized communications.
• It can also bolster in-house policy analysis by synthesizing CRS reports, white papers, and legislative language into bullet points in a matter of seconds.
• Appropriators can get quick summaries of all funding requests and committee reports.
• It can also be helpful for casework, too, including by turning casework into another form of data analytics.

Imagine the positive impact AI can have to improve public oversight with its ability to quickly structure and analyze data about lobbying and the activities of the Executive branch. 

We’re less bullish on more complex tasks like drafting legislation, because the technology is not yet suited for highly-specialized, niche demands where specific words have legal effect, but we are curious to see what develops. Furthermore, AI won’t solve all the problems in minimizing staffers’ workloads. Nor is it a great substitute for other pre-existing tools like Govtrack.us. 

We’re not pollyannaish about the potential risks. AI lacks context and needs human oversight. It’s a good research assistant, but one that can’t be entirely trusted to report accurately. It has no ethics, can fabricate facts, and perpetuate biases. 

Congress also will have to understand how AI may change the lobbying landscape, particularly as tools make it harder to distinguish between authentic and spoofed communications and much easier to generate personalized communications that can overwhelm offices. 

Furthermore, AI platform developers must prioritize security and transparency. Particularly in the congressional setting, it is important to build in mechanisms and standards to protect confidential and sensitive information from becoming incorporated into the machine-learning backend. 

Indeed, smart implementation of AI in Congress will require the development of Congress-specific best practices and tools that will require funding and expertise to develop. 

The House Digital Services has already instituted a working group to experiment with new AI platforms and to assess how to smartly incorporate them into congressional operations. In combination with the new permanent Subcommittee on Modernization within the Committee on House Administration that we encouraged, Congress is much better prepared for AI than it was for the commercial internet when it revolutionized communication technology and the marketplace in the early 21st century.

We’re urging congressional leadership and appropriators to ensure there is institutional support to adequately fund the entities doing this type of work. They need to have enough technologists, developers, dedicated staff, and appropriate sandboxes to experiment and work through the issues AI will create. Likewise, we caution budget hawks to keep funding for congressional staff and not use AI’s ability to improve member office efficiency as a justification for cuts. We’re also fostering  collaboration across support offices and support agencies for the House and Senate, and are making the case for the value for all of these entities to work with outside groups.

Congress must keep pace with this rapidly changing technological landscape. Even the House Select Committee on Modernization barely addressed AI in its final recommendations before sunsetting at the end of the 117th Congress. That had less to do with the ModCom’s capabilities and more indicative of just how fast AI is changing the nature of work and communications. 

If you’re interested in learning more and supporting our efforts to help the First Branch stay current on AI, please get in touch.