Rep. Nikki Budzinski, D-Ill., said some VA employees report claims processing technologies are producing incorrect information. #
Artificial intelligence tools the Department of Veterans Affairs is using to speed up disability claims processing will not replace human reviewers, agency officials told lawmakers Monday, but Democrats contend claims processing also need sufficient staffing levels, not just new technology.
“Every disability claim is decided by a trained VA employee, not by AI or automation. These tools support human decision making; they do not replace it,” Robert Orifici, acting deputy chief information officer at VA’s Office of Information and Technology, said during a House Veterans’ Affairs Technology Modernization Subcommittee hearing.
“What we really strive to do is look at those really high volume, highly administrative tasks, focus our resources and [apply] really basic automation on those capabilities, so that we are really supporting and augmenting the productivity of our employees,” added Derek Herbert, acting chief production officer at VA’s Veterans Benefits Administration.
Rep. Tom Barrett, R-Mich., who chairs the subcommittee, expressed support for VA’s focus on maintaining human disability claim reviewers, saying Congress “will make sure that there are strong guardrails where no veteran claim is fully decided by AI or automation.”
The comments come as VA is actively using AI to speed up disability claims processing by prepopulating toxic exposure memos veterans must file, streamlining information gathering and facilitating other procedures. VA is also planning to acquire or develop additional AI and automated tools for disability claims, Nextgov/FCW reported in March.
But Rep. Nikki Budzinski, D-Ill., the panel's ranking member, argued that the new technologies may actually hinder VA reviewers’ ability to quickly process claims.
“I hear from employees that the so-called fixes that VA has pushed, like automation and pilots of artificial intelligence tools, often produce incorrect information, compounding issues and slowing production,” Budzinski said.
The congresswoman said new technology cannot solve declines in VA staffing, adding that she is alarmed that VA has lost 1100 claims examiners this fiscal year, along with 2700 examiners and 120 IT specialists since January 2025.
Rep. Maggie Goodlander, D-N.H., said she shared Budzinski’s concerns that a lack of adequate VA staffing is leading to less accurate reviews of claims.
Goodlander pressed Hebert to reveal how many claims examiners the VA has on staff and confirm the total has declined since January 2025. Hebert did not have specific numbers but said he would get those to the lawmakers while noting that VA is currently hiring. VA had several open claims examiners positions on usajobs.gov as of Tuesday morning.
Sterling Thomas, the Government Accountability Office’s chief scientist, similarly told lawmakers that federal agencies need a workforce with the right skills to harness new technologies. He argued that solutions such as a federal digital services academy could help VA and other agencies field the talent they need.
Thomas also recommended that VA turn to GAO’s AI framework of key practices as it continues to roll out new AI and automated technologies for claims processing. The framework says federal agencies should document the data used to build AI models, articulate system-level technical specifications, continuously monitor AI performance, assess that performance and take other measures to ensure technology is responsibly deployed.