Witnesses emphasized the importance of the human element in healthcare, particularly in the context of AI. They noted that AI should be considered augmented intelligence, with clinicians remaining in the loop for decision-making.

The VA has adopted a trustworthy AI framework that outlines six principles to ensure AI tools are purposeful, effective, safe, secure, fair, transparent, and accountable. The VA also established the National Artificial Intelligence Institute to promote AI research and development within the VA.

The VA is testing AI technologies that could assist clinical staff with documentation and information extraction, potentially reducing provider burnout and enabling more meaningful interactions with veterans.

The VA believes it is at the middle of the pack in AI adoption compared to private healthcare and teaching hospitals. However, the VA is taking a measured approach to ensure safety and ethical considerations are prioritized.

The VA is in the process of migrating its Vista instances to use the DoD's Oracle Health product, previously known as Cerner. The integration of AI into this new system raises concerns about complexity and potential disruptions to clinical workflows.

The VA is exploring AI applications that could address the unique challenges faced by rural veterans, such as improving access to specialty services through telehealth.

The VA is working on developing procedures and standards to ensure veterans are informed about the use of AI and have the opportunity to provide consent. This is an ongoing concern for the committee, as the VA is already utilizing AI without full disclosure.

The VA emphasizes the importance of protecting veteran data and ensuring it is not monetized by private companies. Existing contracts with vendors include provisions to limit data usage.

Contracting for AI technologies within the government presents challenges, including compliance with FedRAMP for cloud services. The VA is exploring alternative contracting methods to adapt to the rapid pace of AI development.

The VA is utilizing AI models to predict veterans at risk for suicide and provide targeted interventions. The committee expressed concern about ensuring timely clinical follow-up when AI systems flag potential suicide risks.

Witnesses stressed the importance of using high-quality, unbiased data to train AI systems. The VA's unified health record and large patient population offer an opportunity to develop gold standard datasets for AI.

The VA collaborates with academic medical centers, and witnesses noted the potential for partnerships with universities to advance AI research and development.

The committee discussed the possibility of voluntary HIPPA waivers for de-identified data to enable broader AI research and development. Witnesses acknowledged the need to carefully balance privacy concerns with the potential benefits of data sharing.

AI can help VA providers manage capacity and resources by predicting patient demand, optimizing wait times, and improving the allocation of care across direct and community networks.

Members of Congress expressed a need for education and guidance on AI to effectively oversee its development and implementation. Witnesses suggested engaging with industry experts and focusing on specific AI use cases to better understand the technology.