AMA Adopts Policy Pushing Back on AI Creep in Medicine

Artificial intelligence (AI) in medicine must always be overseen by physicians, according to a policy adopted by the American Medical Association (AMA) on Tuesday.

At its annual meeting on Tuesday, AMA’s House of Delegates passed a resolution requiring its leaders to advocate for legislation and regulation requiring AI tools to “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; [and] be subject to relevant and appropriate regulations (including but not limited to those related to liability and documentation).”

The resolution also calls for the AMA to “study emerging concepts around the regulation and licensure of autonomous and semiautonomous augmented and/or artificial intelligence performing clinical functions, and their potential impact on the profession and the physician-patient relationship.”

The AMA must “clearly affirm that the practice of medicine must remain under physician oversight, and that AI should augment, not replace physicians,” Avani Patel, MD, MHA, of Jackson, Mississippi, said Saturday during a discussion of the topic.

She shared a story from another delegate involving an AI program that misread lab results and reported that a patient had an HbA1c of 11.2, resulting in the AI’s diagnosis of diabetes, but the number was actually the patient’s hemoglobin level.

“As AI continues to expand into diagnosis, treatment recommendations, prescribing, and even utilization management, the AMA needs to provide clear policy direction now, while continuing to study emerging issues,” said Patel. “This is a time for physicians to lead, not to wait.”

“This body can make a statement now that AI should not be replacing physicians,” Leanna “Leif” Knight, MD, a Resident and Fellow Section delegate from Providence, Rhode Island, said during the discussion. “While AI may serve as a useful clinical tool, I’m opposed to it replacing physicians with independent AI prescribers or diagnostic systems.”

“We have all made huge sacrifices to become a physician, to be able to be at a patient’s bedside when they’re in need,” she continued. “While AI may one day be more efficient than us, we cannot sacrifice sufficiency for the humanity of being someone’s doctor.”

Knight also warned that AI technologies may “be disproportionately deployed in underserved communities, creating a system where vulnerable patients get algorithmic care instead of direct physician evaluation. AI should automate decisions [but] it should not replace us. We should support policies that preserve meaningful physician oversights and protect patients.”

LaTasha Seliby Perkins, MD, of Alexandria, Virginia, spoke for the American Academy of Family Physicians (AAFP) in favor of additional study rather than passing a resolution, calling it a very complicated issue.

“AAFP has an ethical application of AI as a part of our policies. It shows how AI may reduce administrative burden, but current shortcomings make clinical decision-making support inadvisable without physician verification and guidelines,” she said. “AAFP supports using AI where physicians can verify the output.”

Seliby Perkins was one of several speakers expressing concern about Utah’s state-run pilot program that would allow AI-based prescription renewals. AI-issued medication refills “for hypertension, diabetes, depression, anticoagulation, and controlled substances are not merely administrative transactions,” she said.