How AI can support government services

In the latest Partnership for Public Service AI Center for Government® Use Case Showcase panel, we convened three experts from different levels of government to discuss how public servants can use AI to improve the way government agencies coordinate and deliver projects and services to the public.

Joining the AI Center for this conversation were panelists:

  • Dartanion (Dart) Swift-Williams, chief data officer, City of Baltimore, Maryland
  • Beth Martin, UX/CX design subject matter expert, Federal Innovation Council
  • Basil White, former senior informaticist, Office of the Chief Information Officer, U.S. Department of Agriculture

Improving garbage clean-up

Dart Swift-Williams opened the conversation by discussing Baltimore’s use of AI to combat illegal dumping of trash on vacant lots.

As Swift-Williams explained, illegal dumping is more than just a sanitation issue for Baltimore’s residents. It also causes environmental hazards, the blighting of public and private property and contributes to air pollution in city neighborhoods.

With the municipal government receiving a large volume of service requests to clear illegal dumps, Baltimore turned to AI as a tool to enable a more proactive and efficient cleanup process.

The city deployed an AI-powered camera surveillance system that detects potential illegal dumping incidents and automatically alerts city staff when an event is identified. The staff reviews the flagged camera footage and determines the appropriate response, including dispatching cleanup crews and initiating investigations when unlawful activity is suspected.

While still in the early stages of service, Swift-Williams shared his view that this use of AI will contribute to a stronger government response to illegal dumping. “We believe the impact of this work will be more proactive service delivery, that the city can play a bigger role in detecting illegal dumping events and then we can classify [and address] those events properly,” he explained.

Resolving requirement ambiguities

Panelists Beth Martin and Basil White, both with perspectives from federal experience, discussed how AI can help government agencies make sense of complex governance requirements, uncover hidden risks and improve decision-making.

Underpinning Martin’s and White’s use of AI is a desire to make sense of the enormous and often fragmented volume of rules, directives and requirements that guide the operations of the federal government. To achieve this goal, they developed the Integrated Value Network or IVN, a directed graph of the components, requirements and dependencies of these resources.

Martin and White first developed the IVN more than 15 years ago, first as an Excel spreadsheet, then as a Salesforce resource and now as an AI-powered tool. Explaining the process for using AI, White said: “What we do is find a new law, regulation or executive order, put it in the database and we have the AI compare it to the over 10,000 requirements already in [the IVN].” From there, the IVN identifies how these new policy requirements could impact other agency objectives, enhancing traceability and building a deeper understanding of the relationships between governance resources.

Martin and White also shared two techniques for improving the reliability of AI outputs. First, their “expert misinterpreter” technique uses AI to identify missing assumptions, overlooked risks and potential points of failure in an analysis. Second, their “tweedledumming” approach has two AI models critique and fact-check one another, helping uncover hallucinations and produce more evidence-based responses before human review.

These efforts to improve the reliability of AI outputs reflected the importance of human judgment and analysis in their work. “For all of us in this space, AI can help governments see what people might otherwise miss in terms of connections, ambiguity, risks and dependencies while human judgment and accountability remain essential,” Martin said.

Key takeaways for AI’s role in government projects

  • AI tools should serve a public need. All three panelists stressed that successful AI adoption begins with defined constituent needs, not just a simple desire to use AI.
  • Humans complement AI’s work. The panelists highlighted AI’s role in accelerating key tasks while humans retained positions as analysts and judges of AI’s output.
  • Build trust in AI’s use through responsible practices. These practices include rigorous validation, transparency, critical thinking and accountability for human users of AI.

Throughout the conversation, the panelists made it clear that the core opportunity AI offers to government is to enable public servants to make more informed decisions and deliver better outcomes for communities. “If we are able to make our government more efficient through the adoption of this technology,” Swift-Williams said, “then that gives us more capacity to address the many, many, many things in the backlog.”

Watch the full conversation here:


Continuing the conversation 

The AI Center for Government® champions AI innovators across all levels of government. If your agency is working to lead AI well, we would love to hear from you. Join us as we highlight real-world AI use cases and convene public sector leaders from across the country to share tools and insights to lead confidently in the age of AI.

We’re here to help!