© 2026 Radiology Jobs, LLC. All Rights Reserved. • Privacy Policy | Terms & Conditions
The CEO of America's largest public hospital system says AI could replace radiologists. Here's what the data actually says — and what it means for hiring.
Last week, Mitchell Katz, MD, president and CEO of NYC Health + Hospitals, the largest public hospital system in the country, said something that sent shockwaves through the imaging community.
Speaking at a Crain's New York Business panel on March 25, Katz stated that he is prepared to begin replacing radiologists with artificial intelligence for certain diagnostic reads, pending regulatory approval. His specific target: routine screening mammograms and X-rays, where AI would serve as the primary reader. Radiologists would then review only the cases AI flagged as abnormal.
The response from radiologists was immediate and pointed. Mohammed Suhail, MD, of North Coast Imaging called the comments "undeniable proof that confidently uninformed hospital administrators are a danger to patients." Others in the field pushed back just as forcefully, arguing that current AI technology is nowhere near capable of providing independent patient care.
Both reactions are worth hearing. And if you're an imaging employer or a working radiologist right now, you can't afford to tune this conversation out.
It's worth being precise here, because the headline version of this story loses some important nuance. Katz isn't proposing that radiologists disappear. He's proposing a workflow restructuring: AI handles first reads on low-risk screenings; radiologists step in for abnormal cases and second opinions. Fellow panelist David Lubarsky, MD of Westchester Medical Center Health Network noted that his system's AI misses breast cancers in only about 3 out of 10,000 low-risk negative cases and called the technology "actually better than human beings" in that narrow context.
The caveat Katz himself raised is significant: "if we are ready to do the regulatory challenge." New York state currently requires radiologist oversight of imaging studies. He's not describing something that's happening tomorrow. He's describing where he wants to push the regulatory environment to go.
That's a materially different claim than "AI is replacing radiologists." But it's also not nothing.
The AI-in-radiology debate has been cycling through the industry for years. In 2016, deep learning pioneer Geoffrey Hinton predicted radiologists would be obsolete within five years. That prediction was spectacularly wrong. Radiologist demand is at an all-time high, and Hinton himself later walked back the timeline significantly.
But the conversation keeps resurfacing, and there are real structural reasons why.
Imaging volume is growing faster than the radiologist workforce can keep pace with. Fill times for radiology positions average 130 days, and that's across the board, not just in underserved markets. The Neiman Health Policy Institute recently noted that the radiology workforce may be approaching maximum capacity. Meanwhile, there are now over 700 FDA-cleared AI algorithms for medical imaging in clinical use.
Cost pressure is real too. Radiologists are expensive, appropriately so given training, expertise, and liability, and hospital systems operating on tight margins are looking at every line item. When the CEO of an 11-hospital system says publicly that he sees AI as a path to "major savings," that's not an isolated opinion. It's a signal about where institutional thinking is heading.
If you're responsible for radiology hiring, here's the practical read:
The near-term outlook for radiologist demand hasn't changed. Imaging volumes continue to climb. AI tools in clinical use today are workflow augmenters; they triage, flag, and prioritize. They are not replacing radiologists at the diagnostic level in any meaningful, widespread way. The regulatory environment is still firmly in place.
But the conversation about radiologist roles is evolving. The question of what a radiologist's day looks like in five years, and what that means for compensation, volume expectations, and subspecialty demand, is legitimately open. Employers who are thinking about long-term workforce strategy should be paying attention, not to the headlines, but to the underlying data.
The institutions that will navigate this best are those that treat radiologists as strategic partners in thinking through that transition, rather than cost centers to be optimized away.
We don't have a predetermined answer to where this goes. What we do have is a platform built on the belief that the radiology hiring market works better when it's built on real information and honest conversation.
So we're asking directly:
For radiologists: How are you thinking about AI's impact on your career, and what do employers get wrong when they talk about this?
For employers: Is AI changing how you think about staffing, or is this noise? Where are the real workforce pressures you're navigating right now?
Leave a comment below or reach out directly. This is a conversation worth having in the open.
RadiologyJobs is the first intelligent career advocacy platform built exclusively for the medical imaging ecosystem.