Mitchell H. Katz, the CEO of NYC Health + Hospitals, just said something that every radiologist in America heard loud and clear: he is ready to start replacing them with AI. The only reason he has not done it yet? Regulators will not let him.

Speaking at a Crain's New York Business panel, Katz argued that AI can already match or exceed radiologist performance for certain imaging tasks. His system, the largest public hospital network in the country with 11 hospitals and 70+ community health centers serving over a million New Yorkers, would save enormous sums by automating routine radiology reads.

Sandra Scott, CEO of One Brooklyn Health, a smaller safety-net institution, backed him up. When you are running a hospital on razor-thin margins, the math on AI radiology is not subtle. Radiologists are among the highest-paid specialists in medicine, averaging $350,000 to $500,000 annually. If AI can do 60% of what they do, the cost argument is overwhelming.

Radiologists, predictably, are furious. The pushback online has been fierce, with physicians pointing out that radiology involves far more than pattern-matching on images. Clinical context, patient history, subtle anomalies that AI consistently misses in edge cases: these are not software problems. They are judgment calls that keep patients alive.

Here is the uncomfortable truth both sides are dodging. The radiologists are right that AI is not ready to fully replace them. Study after study shows AI excels at specific, well-defined tasks like detecting lung nodules or flagging breast cancer on mammograms, but struggles with the complex, ambiguous cases that actually matter most. The ones where experience and instinct save lives.

But Katz is also right that the trajectory is clear. The FDA has approved over 900 AI medical devices, with radiology leading the pack. The technology is not perfect today. It will be significantly better in two years. And hospital CEOs are watching the cost curve with the patience of people who know the regulatory dam will eventually break.

What makes this story bigger than one hospital system is the context. The Neiman Health Policy Institute just published data showing the radiology workforce has hit maximum capacity. Wait times are up. Burnout is rampant. There literally are not enough radiologists to meet demand. AI is not just a cost play. It is becoming a necessity play.

The real question is not whether AI will replace radiologists. It will replace some of what they do. The question is who decides when, and whether the transition prioritizes patient safety or hospital balance sheets. Right now, a hospital CEO is telling you his answer. The radiologists should be worried less about whether he is right and more about who is making the rules.