After the Trump administration’s Artificial Intelligence Action Plan was released earlier this summer, health-related agencies are finding ways to train their workforce and incorporate AI into regular operations.
Speaking at the Bethesda Health IT Summit 2025 on Tuesday, federal officials shared how their agencies are working to comply with the AI Action Plan’s mandate that all federal agencies should utilize AI whenever possible.
Jesus Caban, the chief data and analytics officer at Defense Health Agency (DHA), said that as AI efforts ramp up, he’s making sure that DHA employees have access to a number of AI and data science related classes.
“I don’t want to force it on them. It gives them an opportunity for them to learn that in order to grow in your career, either in government or outside, they must know data science. They must know the basic programming,” said Caban. “You know how to tokenize things for AI – do rankings.”
He also pointed to a need to train local chief information officers (CIOs) at medical centers, explaining that CIOs operating under DHA make decisions about what technology to deploy within their individual centers and should be educated on AI and cybersecurity.
To help train the different workforces, Caban said that he is working closely with the DHA J-7 Continuing Education Program Office, which is helping to coordinate required data science and AI training alongside other standard training.
“Staff, regardless of their position, they need to be aware of the guide rails, the risk and the responsibilities we need to follow together [to] embrace these AI models,” said Caban.
The AI Action Plan’s mandate has come with its set of challenges, Travis Hoppe, acting chief AI officer at the Centers for Disease Control (CDC), shared, noting that “we don’t have a choice to do this,” about using AI.
While the CDC has been using forms of large language models and AI systems for years, Hoppe said that AI has varied in its levels of use across different components of the agency. To help prepare the workforce on every level for AI use, Hoppe said that in addition to yearly training, the CDC has a strong community of practice.
“Really proud of this one – about 3000 staff are in one [Microsoft] Teams channel, which is like, give or take, 20% of all of CDC is in one Teams channel, and we meet every other week,” said Hoppe, adding that the agency brings industry leaders, intersting projects, and academics that staff can learn from.
“We also have office hours where staff get to actually talk to each other and ask questions, which is incredibly valuable,” said Hoppe. “And we do things like prompt competitions, which sounds really silly, right, but getting staff to see themselves in the moment. And you know, whether or not you’re a data scientist or an AI engineer, you’re building really cool, high-level stuff. And we have lots of those people at CDC.”
Arjuna Swaminathan, CAIO for the Department of Health and Human Services Office of Inspector General, said that using social learning techniques, like those at the CDC, can help efficiently train staff while operating within a resource constrained environment.
“We’ve launched a AI fluency curriculum that is based on crowd sourced information, crowd sourced from our staff, and we lay out a sort of a foundation that the model for this AI fluency curriculum focused on the type of work that our agency does, so people get to learn about AI in the context of the work that they do,” Swaminathan said about the OIG’s AI preparedness training.
The basic exploratory analytics is something that we would like all of our staff to be able to do, instead of just a select few data scientists, so we’re building up that capacity at the very foundational level for all of our staff,” Swaminathan added.