Radiology is not a dying field. It is, by most workforce measures, a field with more demand than it can currently fill. The American College of Radiology reports more than 1,400 open physician positions posted on its career board, and the aging U.S. population is driving imaging volumes higher each year. The real question behind this search is usually about artificial intelligence, and whether AI will replace radiologists. The short answer: AI is changing radiology, but it is not eliminating radiologists.
Why the “Dying Field” Narrative Exists
The concern traces back to a now-famous 2016 prediction by computer scientist Geoffrey Hinton, who said radiologists would be obsolete within five years. That deadline has long passed, and radiologists are busier than ever. But the prediction stuck, and every new AI headline revives it. Medical students hear it from advisors. Pre-med forums repeat it. And it creates real hesitation about entering the specialty.
The fear is understandable. The FDA has authorized over 1,350 AI-enabled medical devices for the U.S. market as of late 2025, and radiology is the single largest category. These tools can flag lung nodules on chest CTs, detect fractures, measure brain abnormalities, triage strokes, and screen mammograms. On paper, it looks like the job description of a radiologist is being automated piece by piece.
What AI Actually Does in Radiology
The gap between what AI can do in a controlled setting and what it does in real clinical practice is enormous. Despite hundreds of FDA-cleared algorithms, only two have received a full billing code from the system that governs how doctors get paid for medical services. One analyzes retinal images for eye disease. The other estimates blood flow through coronary arteries using software. Neither replaces a radiologist’s interpretation.
Most AI tools in radiology function as triage or decision support. They might flag a scan as “likely positive for pulmonary embolism” and push it to the top of a radiologist’s reading list, saving precious minutes in an emergency. They can measure tumor size more consistently than the human eye, or highlight subtle fractures that might otherwise be missed on a busy overnight shift. These tools make radiologists faster and more accurate. They do not read scans independently, sign reports, consult with referring physicians, or take legal responsibility for a diagnosis.
The complexity of real-world radiology also works against full automation. A single CT scan of the abdomen contains hundreds of structures, and the radiologist is expected to catch everything from an incidental thyroid nodule to early signs of liver disease, all while answering the specific clinical question the ordering doctor asked. AI models are typically trained on one narrow task at a time. Combining them into a system that replicates a radiologist’s full scope of judgment remains far off.
The Workforce Shortage Is Real
If radiology were dying, you’d expect a glut of radiologists competing for shrinking demand. The opposite is happening. The number of students matching into diagnostic radiology residency was 1,084 in 2010. By 2023, it was only 1,006 for diagnostic radiology plus another 123 for interventional radiology, with 100% of positions filled. Training programs are not producing enough new radiologists to keep up.
Meanwhile, the population most likely to need imaging is growing fast. The number of Americans over age 65 increased 38.6% between 2010 and 2020. Older adults require far more CT scans, MRIs, mammograms, and ultrasounds than younger patients. Every new cancer screening guideline, every expansion of lung cancer CT screening or breast MRI, adds volume. Radiology groups across the country are struggling to recruit, and many rely on teleradiology services to cover gaps.
Teleradiology Is Expanding Access
One of the biggest structural shifts in radiology isn’t AI replacing jobs. It’s teleradiology redistributing them. Radiologists can now read scans from anywhere with a secure internet connection, and this market is projected to grow by $3.8 billion between 2023 and 2028, at an annual growth rate of roughly 17.5%. Cloud-based image storage and remote collaboration platforms let radiologists work from home, cover rural hospitals from urban centers, or provide overnight reads for facilities in different time zones.
For radiologists, this means more flexibility in where and how they work. For the field overall, it means a single radiologist’s expertise can reach patients who previously had limited access to subspecialty reads. A pediatric neuroradiologist in Boston can interpret a child’s brain MRI taken at a small hospital in Montana. This is growing the reach of radiology, not shrinking it.
Interventional Radiology Is Booming
Diagnostic imaging is only half of the radiology world. Interventional radiology, where physicians use imaging guidance to perform minimally invasive procedures, is one of the fastest-growing areas in medicine. The market for minimally invasive procedures is projected to reach $94.4 billion by 2030. Interventional radiologists treat liver cancer with targeted radiation beads, open blocked arteries, drain abscesses, place feeding tubes, and perform biopsies, all through small incisions guided by real-time imaging.
These procedures are harder to automate than image interpretation. They require manual dexterity, real-time decision-making, and direct patient interaction. As surgical medicine continues shifting toward less invasive approaches, interventional radiology’s role only expands.
The Financial Pressures Are Real
Not everything about radiology’s future is rosy. Medicare reimbursement rates, which influence what private insurers pay as well, have been declining. The 2025 Medicare physician fee schedule cut average payment rates by about 2.9% compared to 2024. This follows years of incremental cuts. Radiologists still earn well above the average physician salary, but the trend means practices must read more studies to maintain the same revenue, which contributes to workload pressure.
That workload pressure shows up in burnout numbers. A 2023 Medscape survey found that 54% of radiologists reported burnout, up from 49% the year before. Among neuroradiologists specifically, burnout rates have been measured as high as 75%. The causes are familiar across medicine: too many studies to read, too little time, increasing administrative burden, and the emotional toll of sustained high-volume work. AI tools that genuinely reduce repetitive tasks could eventually help here, but adoption is still early.
What This Means If You’re Considering Radiology
The radiologists who will thrive in the coming decades are those who integrate AI tools into their workflow rather than competing against them. Think of it like the introduction of GPS for pilots: the technology handles routine navigation, but you still need a trained human making judgment calls, handling unexpected situations, and taking responsibility for outcomes.
Radiology residency positions fill at 100%. Starting salaries remain among the highest in medicine. Job openings far exceed the supply of new graduates. The imaging volume generated by an aging population shows no sign of slowing. And the procedural side of the field is actively expanding into new territory.
The field is changing, without question. Radiologists ten years from now will use different tools than radiologists today, just as radiologists today use different tools than those who held films up to lightboxes. But a field undergoing transformation is not a field in decline. By every measurable indicator, radiology has more work than it can handle and not enough people to do it.