VA's Use of AI in Claims Processing: Efficiency vs. Accuracy Concerns

The Department of Veterans Affairs (VA) has integrated artificial intelligence (AI) to expedite the processing of veterans’ benefits claims. During a recent House Veterans’ Affairs Committee hearing, Margarita Devlin, principal deputy under secretary for benefits at the VA’s Veterans Benefits Administration, reported a 42% reduction in average claim processing time, decreasing from 141 days to 81 days since January 2025. The VA's 2025 inventory identified 367 AI use cases, with 28 focused on government benefits processing, although many remain in the pre-deployment phase. A notable application, Automated Decision Support (ADS), utilizes machine learning to assist claims processors by organizing relevant information, thereby enhancing efficiency without making final decisions.

NextGov

Lawmakers, including Rep. Morgan Luttrell (R-Texas), expressed support for AI's role in improving service delivery to veterans. However, they emphasized the necessity of human oversight to ensure the technology functions correctly. Devlin clarified that ADS does not make decisions or deny claims but aids decision-makers by compiling necessary information for faster processing.

NextGov

Despite the reported efficiencies, Democratic lawmakers raised concerns regarding the accuracy of claims processing. Rep. Maxine Dexter (D-Ore.) highlighted feedback from veteran service officers indicating a perceived increase in errors, despite improved processing times. Dexter referenced a specific claim that appeared to contain AI-generated content, raising questions about the reliability of the information being processed. Rep. Tim Kennedy (D-N.Y.) challenged the VA's reported accuracy rates, noting a discrepancy between the 94% issue-level accuracy and an 83.31% claims-based accuracy, suggesting that speed does not equate to success.

NextGov

In response to these concerns, VA officials maintained that the accuracy rate had reached its highest level in two years at over 94%. Sandra Flint, Deputy Under Secretary for Field Operations, explained the shift to an issue-based quality rate to provide more detailed insights into claim challenges. The VA also announced a reduction in the backlog of veterans awaiting benefits to under 100,000 claims for the first time since 2020, a decline attributed to the COVID-19 pandemic and the implementation of the PACT Act, which expanded benefits for veterans exposed to toxic substances.

NextGov

Looking ahead, the VA plans to broaden the ADS program to encompass additional conditions and claims, thereby increasing the volume of claims eligible for automated processing. The proposed fiscal year 2027 budget under the Trump administration includes a recommendation for $130 million to enhance automation and AI investments in veterans claims processing. This budget aims to improve operational efficiency and support the ethical use of AI technologies in delivering benefits to veterans.

NextGov

• “the average time to process a claim has dropped by 42% since January 2025, from 141 days to just 81 days.” — Margarita Devlin, Principal Deputy Under Secretary for Benefits, VA

NextGov

• “speed does not equal success.” — Rep. Tim Kennedy, D-N.Y.

NextGov

• “we are eagerly behind the injection of AI models into the system so our veterans get what they need in a timely manner.” — Rep. Morgan Luttrell, R-Texas

NextGov

There is a discrepancy regarding the accuracy of claims processing, with the VA reporting a 94% issue-level accuracy but an 83.31% claims-based accuracy. Additionally, concerns about the potential increase in errors linked to AI reliance remain a point of contention among lawmakers.

NextGov