Introduction
Artificial intelligence (AI) is rapidly transforming health and health care, presenting unparalleled opportunities to enhance patient outcomes, clinical workflows, and health system efficiencies. However, along with these immense possibilities come substantial risks that demand careful oversight and rigorous evaluation. The JAMA Summit on Artificial Intelligence convened experts from diverse health care and technology sectors to deliberate on principles and pathways to develop, evaluate, regulate, and monitor AI tools that serve health needs responsibly and effectively.
Broad Scope of AI in Health Care
AI in health and health care encompasses a wide spectrum of applications. Clinical tools include sepsis early warning systems and diabetic retinopathy screening algorithms, which directly impact patient diagnosis and treatment. Consumer-oriented mobile health applications empower individuals to monitor and manage their health. Health systems also leverage AI to optimize business operations such as revenue cycle management and appointment scheduling. Hybrid AI tools assist both administrative and clinical functions, including clinical documentation, diagnosis suggestion, and treatment planning. Many of these technologies have already become widely adopted, particularly in medical imaging, outpatient visit scribing, and mobile health, demonstrating substantial penetration into everyday healthcare practice.
Challenges in Evaluation and Regulation
While adoption is accelerating, the health effects—both beneficial and harmful—of most AI tools remain inadequately quantified. Challenges in evaluation arise because AI’s impact depends heavily on the specific clinical setting, user training, and the human-computer interface design. Moreover, many AI applications currently operate outside the regulatory purview of the US Food and Drug Administration (FDA), limiting mandatory safety and efficacy assessments. Existing standards primarily emphasize safety monitoring and institutional compliance (such as detection of erroneous model outputs termed “hallucinations”), but often fall short of requiring rigorous evidence that these technologies actually improve meaningful health outcomes.
Key Areas for Advancing AI Evaluation and Deployment
The JAMA Summit highlighted four critical priorities necessary to harness AI’s promise in health care fully:
1. Multi-Stakeholder Engagement Throughout the Lifecycle
Effective AI deployment requires the involvement of end users—clinicians, patients, and other health workers—from the earliest design stages. Moreover, collaboration among developers, regulators, and health systems is essential to facilitate thorough, real-world evaluation during tool deployment.
2. Development of Measurement Tools for Evaluation and Monitoring
New methodologies and expertise are needed to enable rapid, robust, and ongoing assessment of AI effectiveness beyond traditional safety checks. Health systems must be empowered to conduct their own or participate in evaluations that quantify impact on patient outcomes and operational metrics.
3. Creation of a Nationally Representative Data Infrastructure
A comprehensive, data-sharing ecosystem spanning diverse populations and care settings would enable generation of generalizable knowledge about AI performance and health effects. Such an environment supports learning health systems and ensures equitable benefit across varied demographics.
4. Alignment of Incentive Structures
Policy levers and market incentives should promote adoption of evaluation standards and encourage development of AI tools that demonstrably improve outcomes. These measures can help drive systemic change toward safer, more effective AI integration.
Implications and Opportunities
AI promises to disrupt all facets of health and health care delivery profoundly. This disruption offers a rare opportunity to address longstanding challenges in health care access, quality, and efficiency. However, realizing this potential equitably and safely demands building an ecosystem capable of generating rapid, robust, and generalizable evidence about AI impact. Careful, collaborative approaches to development, evaluation, and regulation must underpin this effort to ensure AI improves health for all populations without exacerbating disparities or introducing new risks.
Conclusion
The JAMA Summit report underscores that while AI is poised to revolutionize health care, the trajectory of its impact hinges critically on creating systems for comprehensive evaluation and equitable deployment. Stakeholders must foster multi-disciplinary partnerships, innovate methods for ongoing effectiveness assessment, develop inclusive data infrastructures, and realign incentives. Through these coordinated strategies, AI can fulfill its vast promise to enhance patient outcomes, optimize health care delivery, and advance population health safely and fairly.
References
Angus DC, Khera R, Lieu T, Liu V, Ahmad FS, Anderson B, Bhavani SV, Bindman A, Brennan T, Celi LA, Chen F, Cohen IG, Denniston A, Desai S, Embí P, Faisal A, Ferryman K, Gerhart J, Gross M, Hernandez-Boussard T, Howell M, Johnson K, Lee K, Liu X, Lomis K, London AJ, Longhurst CA, Mandl K, McGlynn E, Mello MM, Munoz F, Ohno-Machado L, Ouyang D, Perlis R, Phillips A, Rhew D, Ross JS, Saria S, Schwamm L, Seymour CW, Shah NH, Shah R, Singh K, Solomon M, Spates K, Spector-Bagdady K, Wang T, Gichoya JW, Weinstein J, Wiens J, Bibbins-Domingo K; JAMA Summit on AI. AI, Health, and Health Care Today and Tomorrow: The JAMA Summit Report on Artificial Intelligence. JAMA. 2025 Oct 13. doi: 10.1001/jama.2025.18490. Epub ahead of print. PMID: 41082366.
