Agencies now have more than 1,700 uses cases for artificial intelligence. The latest data from the Office of Management and Budget shows a more than 100% increase in the number of ways agencies are using these emerging capabilities.
The departments of Veterans Affairs, Homeland Security and Interior are among the agencies with the largest numbers of uses cases.
But the numbers alone don’t demonstrate the impact and benefits of AI to agency missions.
For example, agencies are using AI to analyze drone photos and large datasets. NASA is using AI to implement a global surveillance program to study volcanoes. And there are dozens of examples of robotics process automation and lower-level AI tools that are reducing burdens and accelerating timelines of projects.
Many agencies are still in what the Defense Advanced Research Projects Agency described as the second wave of AI, which is based on machine and/or statistical learning.
Wole Moses, the chief AI officer for civilian at Microsoft Federal, said agencies are using both traditional AI capabilities and are starting to understand both the benefits and opportunities to use generative AI.
“AI technologies have been in play for a while, but generative AI offers some new capabilities, especially around the ability to understand unstructured data, such as at the Social Security Administration with call center conversations,” Moses said on Carahsoft’s Innovation in Government show. “If you look across other agencies’ use cases, think about all the documents and all the research documents, all the policy documents and all the other unstructured data that it would have taken a human previously to understand. That’s an example where technology has enabled new capabilities and the ability to do things faster in more efficient ways, as well as to a better degree of certainty in terms of the outputs.”
AI helps with knowledge management
Moses highlighted the work SSA is doing to use AI to make better sense of unstructured data from citizens and customers calling for help.
SSA used AI tools to reduce call center wait times, which solved one of the biggest frustrations citizens faced.
Moses said SSA needed to gain solid insight into why customers were calling in and what problems they were trying to solve.
“What they did is they used AI to, first of all, transcribe the tens of thousands of calls from voice to text. Then they used generative AI to actually understand what was happening in these thousands of conversations. By doing that, they were able to finally understand some of what was happening in these calls, what are some of the top reasons folks were calling in and then make decisions and changes to some of their internal processes to address some of those,” he said. “This is something that they couldn’t do previously because just the person hours it would take for someone to actually listen into all those calls is something that really is impossible. That’s one customer service example. We’re also seeing agencies use AI chatbots on their external web pages to provide really good interactions with citizens 24/7.”
Along with improving customer service, Moses said another common use case is around knowledge management inside an agency or organization. AI can ease the burden of transferring information from one office or one person to another.
Moses said one common challenge most organizations face is figuring out where specific expertise sits and how to connect information to people.
“Generative AI really flourishes in enabling me to use natural language to find exactly what I’m looking for,” he said. “This is another really powerful example of how AI is really coming to the fore in terms of solving some of these problems that we’ve had for a while that we really didn’t have good solutions to.”
AI innovations coming soon
Moses said as agencies get more comfortable with governing and managing the use of GenAI, the opportunity for innovations using the technology increases.
He said that includes everything from Microsoft 365 Copilot, which became available for government agencies in December, to AI agents and autonomous agents, to different ways to interact with AI, including voice and video.
“Copilot is an AI assistant for your Microsoft 365 environment. It has access to the things that you have access to, but nothing else. It really operates as that personal assistant for you to help you do things like triage your emails much more quickly or help you get started in your day in terms of looking at all of your communications from lots of different sources,” he said. “We get communications in Teams chats, Teams channels and it’s really difficult to keep up across all those things. Copilot can look across all those inbound communications and letting you know what you need to prioritize.”
Moses said with all the excitement and possibilities around AI, there are some common factors that can ensure an agency’s success in implementing these tools.
Like with most technology programs, he said executive sponsorship is probably most important.
In the SSA example, for instance, former Commissioner Martin O’Malley strongly pushed for new tools and technologies, including AI, to improve how the agency served its customers.
A second key factor is a well-developed use case tied to mission or business goals. The old, “don’t use technology for the sake of technology” is even more important when using AI, Moses said.
“The other thing that we’re seeing as one of the things that is driving success is that agencies that really elicit feedback and the voice and support from diverse organizations,” he said. “This shouldn’t just necessarily be a technical decision. We need to have the technical people involved. We need to have end user involved. We need to have the business and the mission involved, and where we’re seeing all of that happen, those confluence of voices and the influence of ideas and feedback really lead to greater success.”
Like with any new technology, Moses said it’s important for agencies to begin testing and piloting AI capabilities to understand the real and potential uses.
“One of the things that we’ve seen some agencies do to be successful is to create communities of interest around AI, where they create opportunities for learning, for sharing and for asking questions,” he said. “We should ensure that those communities have a wide composition of roles, whether it’s researchers, where it’s technical people, business people and mission leaders.”
Listen to the full discussion:
Copyright
© 2025 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
