Every night, my husband and I tell each other about our workdays over dinner. Increasingly, my reports have become a laundry list of AI-related topics. As the Urban Institute’s chief data scientist, my days feature AI-related requests for proposals, evaluation results of generative AI tools, meetings about Urban’s responsible agentic AI strategy, trainings on AI programming tools—the list goes on and on. (Bless my patient and intellectually curious husband!)

I know that I’m not alone. AI, particularly generative AI, has become the primary focus for many of my peers who work at the intersection of technology and public policy—and with good reason! AI has the potential to transform how public policy is developed and delivered.

AI is increasingly becoming the filter through which people find and absorb evidence, including policy research like Urban’s. Government officials at all levels want to know how AI can help them stretch shrinking budgets to cover the delivery of government services, while policy research organizations are asking (and being asked by our funders) how we can use AI to get policy insights in the hands of those making tough decisions.

Already, I know many people are putting their heads down and doing the difficult and thoughtful work to answer these questions. How can we responsibly channel AI’s power in public policy to put evidence in changemakers’ hands and to allow those changemakers to leverage AI for public services?

I am lucky to work with many of these thoughtful people at the Urban Institute, and I am constantly inspired by the commitment to rigor and fairness they bring to their AI work. And we, in turn, have gained so much from opportunities to learn from our peers, who are grappling with the same questions and trying to keep pace as the technology evolves.

To create a space for the field to share practical, evidence-based learning, we decided to start a special vertical within Urban’s Data@Urban platform. This vertical will include quick bites to share what we’re learning as we incorporate AI in our policy work at the Urban Institute and as we support government and nonprofit partners to use AI responsibly in their work. Examples of topics you can expect to see include:

• what we’re learning about making policy data and evidence discoverable to popular AI tools;
• how we’re thinking about evaluating AI systems in policy contexts, including how to power rigorous, human-in-the-loop evaluation at scale by thoughtfully bringing together subject matter expert evaluation with automated and LLM-as-judge approaches; and
• what we’re hearing from our government partners on the ground about questions around AI and how policy research organizations can be good partners.

We’re launching this vertical as I prepare to head out on maternity leave. Soon my dinner table conversations will be less about AI and more about diaper changes and sleep schedules—a temporary change of pace that I’m looking forward to. While I’m out, I am excited for you to hear from my brilliant colleagues. I encourage you to subscribe to Data@Urban if you aren’t already and to join the conversation to share what you are learning. We look forward to learning together.