Nurses Sound the Alarm as a Billionaire AI Investor Pushes AI Care Without Doctors

Part of Nurse.org’s Nursing AI Watch, our ongoing investigation into how artificial intelligence is changing nursing work and who is deciding how it gets used.

It’s time to let AI be the doctors.

At least, that’s what two of the most influential voices in American health policy have just argued, stating that patients might be better off when artificial intelligence, not a human clinician, makes the final call.

In an essay published in The Atlantic on September 8, 2026, University of Pennsylvania bioethicist Ezekiel J. Emanuel and venture capitalist Vinod Khosla wrote that professional medical societies are wrong to insist humans always hold final authority over clinical decisions.

“For Americans to get the best health care possible,” they wrote, “we need to be willing to experiment with models that put AI in control.”

Physician groups have been pushing back since the pair made a version of this argument in JAMA in August. Nurses, who were not mentioned once in the essay, are watching a debate about who gets to practice medicine unfold while their own colleagues are already being laid off and replaced by software.

Want to see more Nurse.org articles in your Google results? Add us as a preferred source.

The argument for AI to take the place of doctors didn’t come out of nowhere. The Atlantic essay expands on an August 17 JAMA Perspective, titled “Will Autonomous AI Exceed AI-Aided Physicians as the Best Medical Care?”

That piece argued that AI already rivals or outperforms doctors on five cognitive medical tasks:

• Gathering patient information
• Choosing tests
• Making diagnoses
• Prescribing treatments
• Managing chronic disease

Its central claim, as quoted by Digital Health Wire, is that “AI alone will provide better medical care than physicians, or even physicians working with AI.” The pair argued that much of medicine is capable of being autonomized, and that human doctors themselves can come across as “robotic,” citing a Gallup survey in which 21% of respondents said they had turned to AI after feeling dismissed by a healthcare provider.

The authors say studies indicate up to 15% of American patients are misdiagnosed each year, leading to tens of thousands of preventable deaths. They also point to a small 2023 randomized study of 32 patients in which autonomous AI got people with type 2 diabetes to a stable insulin dose faster than standard physician-managed care, with better adherence and less diabetes-related distress.

The evidence has caveats the essay’s headline claims skate past. The diabetes trial ran eight weeks, and three of its authors disclosed stock in a startup built on related technology. In addition, Khosla disclosed in the piece that he invested in a virtual primary care startup, co-founded by his son, in which physicians supervise AI assistants. That son, Neal Khosla, CEO of AI virtual care company Curai Health, was also a co-author on the JAMA piece.

The authors acknowledge counter-evidence: in 9 of the 51 studies they reviewed, human physicians outperformed the machines, though they note some of those studies tested older or general-purpose AI models.

Medical organizations were not sold on the pair’s claims. John Whyte, CEO of the American Medical Association, told The Atlantic that he was “disappointed” in the quality of the analysis, questioning study designs built on actors and retrospective chart reviews rather than real patients followed over years. Whyte was open to studying physician-controlled AI tools but only cautiously so on autonomous AI, arguing it would need to be highly regulated.

The AMA’s own position, laid out in an August 18 framework developed with the Digital Medicine Society, identifies five “enduring responsibilities” physicians must retain as AI expands, including preserving human connection, demonstrating clinical judgment, and stewarding safe AI integration.

The American College of Physicians wrote in a recent position paper that “fully autonomous AI is possible,” but cautioned that its use should be infrequent and “limited to low-risk, low-complexity decisions, and always allow for bringing a clinician into the loop.”

>>Listen to The Latest Nurse News Podcast

While physicians debate a hypothetical world of being replaced by AI, nursing is facing it.

In July 2026, the New York State Nurses Association warned that “Montefiore is planning to lay off real nurses and replace them with artificial intelligence (AI)-powered software,” referring to RNs in the Bronx who used bedside experience to fight insurance denials on patients’ behalf. Nurse.org covered those layoffs as they unfolded.

In August, STAT reported that nurses are pressing for a seat at the table as clinical AI expands, citing both the Montefiore layoffs and workplace surveillance concerns at Kaiser Permanente.

National Nurses United, which represents more than 200,000 RNs, has staked out a harder line than any physician group: “Nurses are not against scientific or technological advancement, but we will not accept algorithms replacing the expertise, experience, holistic, and hands-on approach we bring to patient care.”

The trust gap shows up in the data. In a 2024 NNU survey of more than 2,300 registered nurses, 60 percent said they did not trust their employer to put patient safety first when implementing AI, and 69 percent of those working with acuity algorithms said the output did not match their own assessment. Elsevier’s Clinician of the Future 2026 nurses report found that only 42% of nurses worldwide consider AI tools trustworthy today, and just 41% use AI for work compared with 57% of doctors.

Nothing in the Emanuel and Khosla proposal is law, and no regulator has authorized an AI to practice medicine independently in place of a physician. The FDA has cleared narrow autonomous diagnostic tools, like software that screens for diabetic retinopathy without a clinician reading the result, but nothing close to a doctor’s full scope.

As a nurse myself, I feel like there are two sides to the stance. On one hand, it feels like a slippery slope to exclude people from healthcare. On the other, if the doctor side of the medical system becomes AI-augmented, would that open up more respect, time, and pay for the hands-on clinical work of nursing?

But the framing matters, because arguments that start with physicians rarely stop there. If the standard becomes whether an algorithm beats the median clinician, nursing judgment, which is harder to measure in a simulated encounter, is the easier target for automation. And nursing care is still bundled into hospital room-and-board charges rather than separately reimbursed, an obstacle that needs tackling before we talk respect in dollars.

Still, nurses aren’t waiting around for a decision. The nursing profession’s own institutions are moving.

On May 5, 2026, the American Nurses Association called for nurse-led guardrails on AI, publishing consensus findings from its April 2026 AI in Nursing Practice Think Tank that flagged erosion of professional judgment, unclear liability when AI influences a decision, algorithmic bias, and the absence of nursing-specific governance frameworks. “Artificial intelligence is already shaping nursing work in critical ways,” said Brad Goettl, DNP, DHA, APRN, FAAN, FACHE, chief nursing officer of the American Nurses Enterprise, adding: “The profession is at a pivotal moment that requires deliberate, nurse-led action to protect patient safety, ensure nurse well-being, and sustain public trust.”

Practically, that means three things for bedside nurses right now:

1. Know whether your employer has an AI governance committee and whether a nurse sits on it. Among the 31 health-system AI governance structures Nurse.org could document in detail, only six show a named nurse leader publicly shaping AI governance.
2. Document every time an algorithm’s acuity score or alert conflicts with your assessment, because that paper trail is what regulators and arbitrators rely on.
3. Read your contract language on technology, since union agreements are currently the strongest tool nurses have to require notice before AI is deployed on their units.

Related Nursing AI Watch Analysis

🤔 If an AI tool overruled your clinical assessment tomorrow, would your unit have a clear process for escalating it, or would you be on your own? Tell us in the comments.