Government agencies are increasingly turning to artificial intelligence (AI) to enhance workforce capacity, improve mission effectiveness and drive operational efficiencies. In a July 30 FedInsider webinar, Federal IT leaders discussed the “building blocks” of AI in the Federal government.
Keith Nakasone, deputy assistant commissioner of IT Acquisition at the General Services Administration (GSA), stressed the importance of actually understanding the technologies before rushing to deploy them. He explained that GSA is working to understand AI, along with other emerging technologies, so it can develop an agency-wide strategy. In terms of how GSA is leveraging AI for the other Federal agencies it works with, Nakasone said it is using the knowledge it is gaining to enable rapid AI deployment through existing and future contract vehicles.
Ted Kaouk, chief data officer for the Department of Agriculture (USDA), explained that the USDA is trying to “weave the artificial intelligence and machine learning [ML] technologies into the heart of our IT modernization activities as a whole.” As it deploys AI and ML, USDA is looking to answer three primary questions:
- How can AI and ML enable employees to focus on higher-value work?
- How can the agency use predictive modeling to be able to anticipate and mitigate risks in its program areas?
- How can AI and ML improve the customer experience?
Brett McMillen, general manager, U.S. Federal Civilian and Ground Station, Amazon Web Services, explained that previously, due to cost, deploying AI and ML was limited to only the largest agencies. However, there is now a push to “democratize” the deployment of these emerging technologies across all agencies.
Tod Dabolt, director of information management at the Department of the Interior, delved into how his agency is using machine reasoning, which he called “rather cutting-edge” for the agency. He explained that by structuring open data in a way that is more understandable to machines, the agency hopes to improve the customer experience. Dabolt said that goal is to prevent users from having to wade through piles of open data to find the answer they are looking for by having machines that can understand their question, reason against it, and find the right answer.
In terms of how agencies can begin deploying AI, Kaouk urged IT teams to work with their agency leadership to understand what are the “key questions they are trying to answer.” Dabolt agreed with Kaouk and explained, “having a one-off AI project to answer a specific question is one thing. Having the capability where your data is organized and cataloged in a way that you can take advantage of it through multiple of scale with a greater economy of scale is what we’re driving towards.”
