AI Use Must Align with Organizational Strategy, VA AI Lead Says

The Department of Veterans Affairs is aligning its artificial intelligence strategy with existing agency priorities – like reducing physician burnout and improving clinical outcomes – to innovate with a purpose and increase literacy before fully integrating AI into its operations.

“We are not offering any directly veteran facing AI tools at this time, just because we’re not sure about the risk levels. … Our strategy there is to provide the tools to VA employees who are that human in the loop and make sure that the right information gets to veterans,” Dr. Kaeli Yuen, data and AI health product lead at VA’s Office of the CTO explained during the Google Public Sector Summit Oct. 16.

VA has been reviewing the feasibility of pilots to test technologies that record conversations or synthesize data for clinicians so they can spend more time caring for patients and less time doing administrative work.

This May, VA hosted an AI tech sprint competition where the agency selected ten winners to pilot their technology to alleviate clinician burnout. According to Yuen, those pilots will hopefully begin by the end of this year.

As VA develops new AI solutions, the agency is relying on “clinical champions” to test, train and provide feedback on the tools.

“The success of any clinical pilot is highly dependent on having an engaged clinical champion,” Yuen said. She said the VA looks for a “very engaged clinical champions who would serve as a user, who would dedicate time to teaching other physicians, who would dedicate time to providing feedback about the tool and how to incorporate that into the workflow.”

Overall, exposure to the technology – whether through pilots or clinical champions – serves VA’s larger purpose of increasing employees’ literacy around AI.

“Obviously there’s the traditional things like trainings, but I’ve found that at this point, exposure is quite helpful, even if it’s a simple tool,” Yuen said.

She pointed to an internal generative AI tool in the pilot stage that employees have used to get familiar with the technology.

“Through that exposure, VA employees were able to actually teach us a lot about potential use cases for these tools. They gained exposure and gained familiarity with the tools … and started to feel a lot more comfortable with AI tools in their everyday life,” Yuen said.

In the long term, Yuen believes VA will embed generative AI fully into its operations. In the meantime, the agency is focused on increasing employee literacy and comfort with the technology.

“Our perspective is that in the not-too-distant future, generative AI is going to be embedded into most of the enterprise software that we use. However, as we’re waiting for that to happen, it is still worth it to build some things in-house, to learn, to build the capacity amongst employees to use these types of tools,” Yuen said.