The Timing Couldn’t be Better

Right now, the spotlight is on the intersection of artificial intelligence and healthcare, but for all the wrong reasons. Recently, an Anthropic safety researcher left the company, citing its reckless behavior, while another raised a stark warning that AI systems built today might carry a 10% chance of leading to human extinction within a decade. Amidst this alarming backdrop, Dr. Ezekiel Emanuel is pushing for AI to be our primary healthcare provider.

As someone who has been in internal medicine for nearly 30 years, I get where Emanuel is coming from. His enthusiasm for AI in medicine is commendable, I suppose. However, I believe he’s quite mistaken.

When it comes to the profound moments in our lives—those times of joy or despair—what we fundamentally need is another person alongside us.

Let’s start by discussing Emanuel’s recent essay in the Journal of the American Medical Association, one of the leading journals in the field. He cites various “studies” that suggest AI-only care could outperform care from human physicians or even a mix of both. He lists five areas where he feels AI excels: gathering relevant medical information, determining diagnoses, outlining diagnostic tests, prescribing treatments, and managing chronic issues.

But there’s a catch—details matter.

To start with, none of the studies Emanuel references meet the rigorous standards most would expect: thorough randomized, double-blind trials with substantial clinical endpoints. Instead, he appears to lean on small observational studies and simulations, none of which can definitively conclude anything significant.

If anything, these publications raise more questions than they answer. Emanuel seems to suggest that we should conduct more research. Fair enough, but in a subsequent piece for The Atlantic Monthly, he accuses cautious medical professionals of ignoring the evidence available and potentially hindering public health.

That leads to my question for Emanuel: with your reliance on its questionable studies, what credible evidence are you really presenting?

Additionally, it seems Emanuel is quite taken with the idea of algorithms. While established clinical pathways can guide symptoms and treatments, seasoned physicians recognize that clinical judgment is not something that can simply be programmed into a computer.

Patients and their issues are never just numbers or frameworks. Daily, doctors face complexities and unexpected situations. What’s needed goes beyond textbooks and algorithms; we rely on intuition, open-mindedness, and a kind of practical wisdom that comes only with experience.

Emanuel’s perspective reminds me of rationalists criticized by philosopher Michael Oakeshott—those who idolize technical knowledge found in textbooks while dismissing the essential know-how gained through lived experiences. It seems appealing in theory, but when you put it into practice, it tends to falter.

There’s also his strong faith in consensus statements and clinical guidelines, which are often suggestions rather than firm evidence. The reality is that much of this guidance is only moderately reliable at best. As Abba Eban once remarked, “A consensus means that everyone agrees to say collectively what no one believes individually.” Although AI can access a vast amount of data, human physicians often imbue that information with wisdom, considering nuances with phrases like, “It depends.”

If Emanuel argues that human decisions are marred by error, it’s worth noting that it’s human errors that largely inform AI systems. Ironically, the imperfections of humanity get woven into the fabric of artificial intelligence.

At the end of the day, I think he undervalues the human connection. Surely, as someone skilled in oncology, he recalls those pivotal moments when he stepped back from his screen, took off his glasses, and genuinely engaged with a patient. Those instances bring the profound connection of the physician-patient relationship into sharp focus.

Are we really led to believe that this relationship doesn’t matter? That it’s merely a trivial aspect of navigating sickness and health? Dr. William Osler, a pivotal figure in medicine, would surely challenge such a viewpoint. He once asserted, “The good doctor treats the disease; the great doctor treats the patient who has the disease.”

Perhaps research will yield different results down the line. Perhaps, one day, AI could indeed surpass human physicians (or those aided by AI), and we might find ourselves accepting our computer counterparts as they dispense medical care with cool precision.

But honestly, I’m not buying it.

There’s a poignant story about a man who wept in a crowd when news broke of President Franklin Roosevelt’s passing in 1945. When asked if he knew Roosevelt personally, he simply replied, “No, I didn’t know President Roosevelt, but he knew me.”

In our highest and lowest moments, we will always desire the presence of another—someone who truly understands us, who cares, and who knows us well.

AI is indeed a powerful tool, yet it doesn’t truly know who we are. So, find a physician who does.