As augmented intelligence (AI) deepens its roots in medicine, delegates to the 2026 AMA Annual Meeting moved to ensure that evidence-based care is integrated in all health AI systems and that transparency remains at the forefront of efforts.
One big area of concern: The way in which AI-enabled decision support systems are reshaping how information is synthesized and applied in the practice of medicine. AI tools can offer rapid analysis and evidence-informed recommendations that enhance human expertise, not replace it. And they can improve efficiency, workflow and detection of high-risk conditions. But there is reason to be wary.
“AI has enormous potential in healthcare, but it cannot replace physician judgment,” said AMA CEO John Whyte, MD, MPH. “Patients deserve care decisions that are informed by the latest medical evidence and guided by a physician who understands their individual needs. Whether AI is helping a physician make a clinical decision or assisting with an insurance review, there must always be transparency, accountability, and meaningful physician oversight. Technology should support better care—not stand between patients and the care they need.”
Long-term validation of AI’s full impact on patient outcomes is undetermined and concerns about transparency, bias and explainability linger, says an AMA Council on Science and Public Health report introduced at the meeting.
Today, evidence-based medicine is the cornerstone of clinical decision making in medical practice, using a multistep process to provide the best outcome for each patient. Integrating those principles—such as graded evidence hierarchies and rigorous appraisal—into AI systems “could strengthen confidence in integrating AI-enabled technologies into clinical care,” the council’s report says.
“However, significant challenges persist, including opaque model reasoning, inconsistent standards for transparency, a lack of transparency mandates in regulatory structures, and risks of confabulations in generative models,” the report says. “As clinical guidelines and medical education continue evolving to keep pace with rapid evidence generation, AI offers substantial potential to support real-time updates and organize and analyze large amounts of information, provided that transparency, reliability and user-centered design remain central.”
The AMA says that “AI should be viewed as an augmenting tool that complements human judgment, reinforces evidenced-based practice and supports clinicians in delivering high-quality, patient-centered care.”
To foster evidence-based medicine in AI tools, the AMA will:
- Recognize and promote the importance of transparency and explainability so physicians have sufficient information to make sound clinical decisions when using AI clinical decision support tools, which depending on the tool, may include information such as the grading of medical evidence including the data sources.
- Collaborate with medical specialty societies, relevant key parties, regulators and AI developers to establish standards and develop a framework for evidence attribution, evaluation and validation in AI clinical decision support systems.
- Encourage medical education stakeholders to incorporate training on the utility, limitations and interpretation of evidence-based medicine practices when using AI tools in clinical decision-making.
- Monitor best practices and policies of AI transparency and evidence-based recommendations to improve the quality and reliability of patient care.
From AI implementation to digital health adoption and EHR usability, the AMA is fighting to make technology work for physicians, ensuring that it is an asset to doctors. That includes the AMA Center for Digital Health and AI, which works to ensure physicians help shape how AI is developed, implemented, and regulated across the healthcare system, with patient safety and physician-led care remaining at the center of those efforts.
Taking care with AI-generated clinical notes
In an effort to be more efficient and save time, physicians and other health professionals are beginning to use AI to generate notes. But with that comes physician and clinician responsibilities.
Physicians need to make sure they understand the tool they are using. The medical record could contain an error specifically related to AI-generated documentation that could interfere with care.
Given that risk, the House of Delegates adopted new policy stating that “prior to the use of AI in the medical record, training in the use of AI is highly recommended and to include the benefits of AI, as well as the potential harms that could exist in an AI-generated document.”
AI-driven decision transparency
It’s becoming increasingly common for health plans to use AI and algorithmic black-box tools to deny, reduce, or terminate coverage or payments for medical care. But these tools often rely on datasets that don’t reflect regular updates in the science or clinical guidelines across specialties.
While the AMA’s human-in-the loop policy ensures that a physician makes a final call, the AMA will also boost its efforts in this area by advocating:
- Federal and state regulations and legislation requiring health plans and third-party payers to provide physicians and the insured patient with the specific clinical logic, evidence-based sources and version history of any AI or algorithmic tools used in the issuance of an adverse determination.
- That any AI-driven or algorithmic tool used for clinical review must be transparently audited—with reaudits triggered by material changes to the AI model, its training data, or applicable clinical guidelines, and with periodic comprehensive audits at minimum annually regardless of such changes—to ensure it reflects the current evidence-based clinical guidelines and recognized standards of care.
“When health plans use AI-driven tools to deny or delay care without explaining how those decisions were reached, physicians and patients are left in the dark,” Dr. Whyte said. “AI should never function as an unaccountable black box. Health plans must be transparent about how these tools work, what evidence and data sources they rely on, and whether a qualified physician reviewed the decision.”
Health AI and physician-led, team-based care
With physicians uniquely qualified to lead coordinated care teams—a model that has been shown to be the best way to support high-quality care and lower costs—the AMA will “make efforts to educate physicians on evidence-based artificial intelligence tools that can strengthen collaboration with non-physician clinicians through improved interdisciplinary communication, decision support and workflow integration.”
The House of Delegates also adopted policy to:
- Support the development and dissemination of best practices for AI integration into physician-led care teams, with emphasis on safety monitoring, transparency, cyber hygiene and preserving physician leadership in clinical decision-making.
- Encourage further research on AI interventions that demonstrate how AI can enhance physician-led team effectiveness, reduce misalignment of clinical risk perception, and improve coordination of care.
In a separate action, delegates directed the AMA to advocate legislation and regulation related to the use of AI in autonomous or semiautonomous circumstances in healthcare—including diagnostics, prescriptions, care management or other functions—requiring that such tools must:
- Integrate with the physician-led team and be used at the direction of the treating physician.
- Respect the continuity of care and best practices related to transitions of care.
- Have transparent, auditable data demonstrating safety and efficacy.
- Be subject to relevant and appropriate regulations, including but not limited to those related to liability and documentation.
- Adhere to AMA policy on AI in healthcare (PDF).
Delegates also directed the AMA to “study emerging concepts around the regulation and licensure of autonomous and semiautonomous AI performing clinical functions, and their potential impact on the profession and the patient-physician relationship.”
Read about the other highlights from the 2026 AMA Annual Meeting.
