AMA Disputes JAMA Paper Claiming AI Outperforms Human Doctors

The American Medical Association is pushing back against a provocative paper published in the Journal of the American Medical Association (JAMA) claiming that artificial intelligence is already surpassing human physicians in key cognitive medical functions. The dispute centers on whether AI can provide superior care independently or if human oversight remains an essential safeguard for patient safety.

The paper, authored by bioethicist Ezekiel Emanuel and venture capitalist Vinod Khosla, asserts that AI-alone medical care is likely to be more effective than care provided by physicians alone or through human-AI hybrids. Specifically, the authors claim that AI surpasses human doctors in five fundamental tasks: eliciting medical information, diagnosing patients, conducting diagnostic testing, and managing chronic diseases.

Methodology Disputes

John Whyte, CEO of the AMA, has challenged the validity of these findings, arguing that the evidence cited in the paper relies heavily on simulations rather than gold-standard blinded experiments. Whyte contends that simulated performance does not necessarily translate to real-world clinical success. To support this, he pointed to a Nature article indicating that patients have encountered significant difficulties when using AI chatbots to obtain the medical expertise they require.

While the AMA acknowledges the potential of these tools, Whyte emphasized that they must be utilized within a care plan governed by a physician. This stance reflects a broader concern regarding the rapid integration of Large Language Models into the medical field since late 2022, as the industry grapples with the risk of "AI-deskilling" among practitioners who may become overly dependent on automated tools.

The Stakes for Healthcare

This conflict highlights a fundamental tension between the drive for diagnostic efficiency and the necessity of clinical judgment. If the authors' premise is correct—that human intervention can actually degrade the performance of AI—it could trigger a paradigm shift in how medical care is delivered and governed, potentially reducing the physician's role to a secondary or administrative one.

However, the AMA maintains that the patient-provider relationship and human oversight are non-negotiable components of safe medicine. The debate now shifts to whether future, more rigorous blinded studies will support the authors' claims or confirm the AMA's skepticism regarding the autonomy of medical AI. The resolution of this conflict will likely define the boundaries of AI integration in clinical settings for the next decade.