AI vs. Radiologists: The Fight Over Who Gets to Read Your X-Ray Is Getting Ugly

Why this matters: If a hospital CEO has his way, the next person reviewing your chest scan won’t be a doctor — it’ll be software. And the people who built that software are making promises that actual physicians say are dangerous, irresponsible, and frankly untrue.

According to a report from Times of India, the CEO of NYC Health + Hospitals — the largest public hospital system in the United States — has stated plainly that he’s ready to start replacing radiologists with AI. He’s not hedging. He’s not saying “supplement.” He said replace. And the tech executives cheering him on are now being called out by the medical community for spreading what doctors are calling misinformation.

This is one of the most important conversations happening right now, and most people aren’t paying attention to it.

What Nvidia and Anthropic’s CEOs Actually Said

Jensen Huang of Nvidia and Dario Amodei of Anthropic have both made bold public statements suggesting AI is reaching — or has already reached — a point where it can perform at or beyond physician-level competence in specialized fields like radiology. Amodei went further, suggesting AI could compress decades of medical progress into just a few years.

That sounds incredible. It also sounds like a pitch.

Doctors aren’t buying it. Radiologists and medical professionals across the country pushed back hard, pointing out that AI diagnostic tools still produce false positives, miss context, and lack the clinical judgment that comes from years of actually treating patients. One radiologist put it bluntly: AI doesn’t know that the patient just had surgery last week. It doesn’t know they’re anxious and hyperventilating. It reads a file. A doctor reads a person.

And yet, hospital administrators are nodding along to the tech CEOs. That’s the part that should make your stomach drop.

The Business Logic Is Obvious — And That’s the Problem

Radiologists are expensive. A seasoned radiologist can earn well over $400,000 annually. AI software, once licensed and deployed, costs a fraction of that. From a pure spreadsheet standpoint, the math is seductive.

But here’s what that spreadsheet doesn’t include: liability when the algorithm misses a tumor. Accountability when a patient dies from a misread scan. The quiet institutional pressure on hospital staff not to override the AI because it slows things down and creates paperwork.

We’ve seen this movie before. Automation replaces workers. Costs drop. Executives celebrate. Then something goes wrong, and suddenly everyone wants a human in the loop again — except there are no experienced humans left because we trained them out of the system.

AI is transforming every corner of the media and tech world too. OpenAI recently acquired a tech industry talk show, which tells you everything about how aggressively these companies are shaping the narrative around their own products. When AI companies own the microphones, skeptics don’t get airtime.

The Irony of “Democratizing” Healthcare

Advocates for AI in medicine love the word democratization. The argument goes: AI can bring expert-level diagnostics to underserved communities that don’t have access to top specialists. Rural hospitals. Low-income clinics. Developing nations.

It’s a compelling argument. And it’s being used to justify replacing well-trained doctors at well-funded urban hospitals that absolutely could afford to keep them.

NYC Health + Hospitals serves a predominantly low-income, minority population. These are patients who already face systemic disadvantages in healthcare. They deserve more human attention, not less. The idea of using them as a testing ground for AI replacement — framed as innovation — is something the medical community should resist loudly.

AI tools are getting smarter, no question. Google’s Gemma 4 model can now handle text, image, and audio tasks simultaneously, and multimodal AI will absolutely have legitimate uses in clinical settings. But “useful tool” and “replacement” are two very different things.

🔥 Hot Take: Tech CEOs Are the Wrong People to Trust on This

Here’s the uncomfortable truth: Jensen Huang and Dario Amodei are not neutral observers. They are selling AI infrastructure and AI models respectively. Every hospital system that adopts their technology is a paying customer. When they say AI can replace radiologists, they are not offering medical advice. They are closing a deal.

The fact that hospital administrators — people with purchasing power and cost-cutting mandates — are listening to tech billionaires over practicing physicians is a structural failure. It’s not innovation. It’s salesmanship wearing a lab coat.

AI will play a role in the future of medicine. That future should be built by doctors, ethicists, and patients — not by executives whose quarterly earnings depend on hospital adoption rates. Until then, be very skeptical of anyone who tells you an algorithm is ready to replace the person looking you in the eye and deciding what’s wrong with you.

That gap — between reading data and understanding a human being — is not a software update. It’s everything.