Recent claims that AI has reduced the need for radiologists are contradicted by evidence showing medical imaging demand is projected to grow 17-27% by 2055, imaging volumes have increased significantly since 2016, and radiologist workload continues to rise despite reimbursement cuts of nearly 50%.
- Imaging growth projection: U.S. medical imaging use expected to grow 17-27% by 2055, driven by population growth and aging
- Volume increases: Medicare Part B data shows significant CT and MRI volume increases between 2016 and 2024
- Workload growth: Imaging relative value units (RVUs) per beneficiary reached highest levels in 2021, up 13% since 2005
- Reimbursement decline: Medicare inflation-adjusted radiologist reimbursement cut 48.3% between 2005 and 2025 despite growing workload
- Repeat imaging concern: 12.5% of Medicare office-based imaging studies were repeated between 2013 and 2022, contradicting claims about EHR reducing duplicates
Recently published comments about AI's impact on the radiology workforce may be misleading, according to an editorial published in the Journal of the American College of Radiology.
The comments, published in 2026 by former White House economic advisor David Cutler, PhD, and Lev Klarnet from Harvard University in JAMA Health Forum, suggested that the decrease in medical imaging has "allowed the U.S. to meet the need for imaging without an increase in radiologists."
"The assertion, we think, is counter to recent collective experience of the radiology community and ignores a decade of evidence," wrote a team led by Elizabeth Rula, PhD, executive director of the Harvey L. Neiman Health Policy Institute (HPI) in Reston, VA.
Medical imaging is no stranger to broad declarations by research and policy luminaries. Nobel laureate Geoffrey Hinton, PhD, in 2016 made his now-infamous comments about deep learning being able to outperform radiologists in imaging analysis.
Since then, he has backtracked. In a 2026 New York Times interview, said a combination of AI and radiologists will make radiologists more efficient while improving accuracy.
Rula et al wrote that the concern "remains newsworthy and radiologists remain defensive, indicating the longstanding ripple effect of Hinton’s prediction."
Cutler is the Otto Eckstein Professor of Applied Economics at Harvard University and the dean of the Social Sciences division of the Faculty of Arts and Sciences.
Cutler and Klarnet's analysis of imaging utilization and policy relies on decade-old data that may not accurately reflect current trends, the opinion authors wrote. The key study they cite examined per capita imaging use through 2016 and attributed part of the apparent decline to Medicare's code bundling, a billing change. The analysis also fails to account for radiologist workload, which is driven by both examination volume and complexity and has continued to grow, Rula and colleagues wrote.
The HPI team cited peer-reviewed studies it wrote contradicts Cutler and Klarnet's conclusions. The team in 2025 published research projecting U.S. imaging use will grow 17% to 27% by 2055 driven by population growth and aging. Medicare data from the Neiman Almanac shows significant Part B volume increases in CT and MRI between 2016 and 2024.
And a separate study by the HPI team, which measured radiologist workload in relative value units (RVUs), found that imaging RVUs per beneficiary reached their highest level in 2021, up 13% since 2005.
The HPI team also cited evidence that challenges Cutler and Klarnet's claim that electronic health records have reduced repeat imaging, with a recent analysis finding that 12.5% of Medicare office-based imaging studies were repeated between 2013 and 2022.
And while reimbursement cuts have been cited as a tool to reduce imaging volume, Medicare's budget neutrality mandate has slashed inflation-adjusted radiologist reimbursement by 48.3% between 2005 and 2025 even as workload has grown, Rula et al wrote.
The HPI authors expressed concern that Cutler's comments "could misleadingly assuage very real concerns and the necessary efforts to ensure appropriate imaging utilization and manage workload."
They added that downplaying challenges of AI in medical imaging could have a ripple effect like Hinton's comments in 2016. This could have a negative impact on patient access to necessary and valuable imaging.
Read the HPI team's full opinion here.
