AI Won’t Replace Doctors, Says AMA CEO: A Human‑Centred Take on the Future of Care

When you get a message from a chatbot that says you might have cancer, does that feel comforting? Not really. Most of us would still want a real doctor to sit with us, look at the scans, and walk us through the next steps.

That very question sits at the heart of a recent debate that I, John Whyte, the chief executive of the American Medical Association, had with oncologist‑bioethicist Ezekiel Emanuel and health‑policy analyst Abe Baker‑Butler. Their point: AI is getting good at tasks that used to be the sole domain of physicians—reading images, summarizing records, even suggesting treatments. My point: good at tasks doesn’t mean good at being a doctor.

Let’s be clear. AI can already spot a nodule on a CT scan faster than most radiologists, can pull together a patient’s entire chart in seconds, and can churn out a draft note that sounds almost human. Those achievements are impressive, and we need rigorous studies to know where the technology shines and where it stumbles.

But the leap from “AI can do a lot of what doctors do” to “AI can replace doctors” is a shaky one. Medicine isn’t a to‑do list. It’s a conversation, a relationship, a series of judgment calls that blend science with compassion. It’s noticing that a patient’s story doesn’t quite match the textbook, or that a guideline‑approved drug would clash with a family’s cultural values. It’s delivering bad news with a steady hand, and knowing when to step back and just listen.

Patients don’t walk into an exam room as neat case studies. They stumble in with vague symptoms, fragmented histories, and a swirl of emotions. Their choices are filtered through fear, hope, finances, and personal priorities. Those messy, human factors are things no benchmark test can capture.

Another piece of the puzzle is responsibility. Doctors are licensed, bound by ethics, and ultimately answerable for the care they provide. If an algorithm misreads a scan or offers a treatment that backfires, who takes the heat? The manufacturer? The hospital? The clinician who trusted the tool? As AI takes on bigger roles, we need clear lines of liability, and we need the companies that build these systems to step up and own the risks.

All of this isn’t a call to halt progress. On the contrary, the tools we’re seeing today can make a real difference. A gastroenterologist armed with AI assistance might catch a tiny polyp that would have slipped by. A radiology department could flag early‑stage lung cancers that were previously missed. And on the administrative side, AI can tidy up paperwork, giving doctors more time for the parts of the job they actually love—talking to patients.

Perhaps the most exciting promise is democratization. Imagine a rural clinic that, through a secure AI platform, can tap into the same diagnostic expertise that a top‑tier academic hospital provides. That could narrow the gap in cancer outcomes, chronic‑disease management, and more.

In short, we should welcome AI as a teammate, not as a replacement. The technology will keep getting smarter, and physicians should be at the table shaping how it’s used, setting the ethical guardrails, and ensuring that the final decision rests with a human who understands the whole person, not just the data points.

So, if a future AI system tells you you have cancer, you’ll still want a physician beside you—to interpret, to explain, and to walk you through the road ahead. That’s the essence of medicine, and it’s not something code can replicate.