November 18, 2026
Me for Him: What Medical Specialties Are Most Vulnerable to Substitution by AI?
You think we look pretty good together
You think my shoes are made of leather
But I’m a substitute for another guy
I look pretty tall, but my heels are high
The simple things you see are all complicated
I look pretty young, but I’m just back-dated, yeah
Substitute, The Who, 1966
If you know me, you know that I look for any opportunity to work in The Who, the greatest rock band of all time. That’s why I couldn’t pass up a new Perspective published earlier this month in NEJM AI. Written by five researchers affiliated with the University of California at Berkeley, Stanford University and the U.S. Census Bureau, the piece used Medicare Part B claims data to project what medical specialties are most susceptible to substitution by artificial intelligence (AI).
The study pool was nearly 600,000 physicians in 56 specialties who received about $45 billion in Part B payments in 2023.
The researchers teased out the payments to the doctors in each specialty by how much was for physical contact with a patient, how much was for non-physical contact with a patient, and how much was for a mix of physical and non-physical contact with a patient. The researchers said AI could be a substitute for the non-physical contact part.
Side note: The researchers’ ability to tease out payments by services performed by medical specialty was made possible by billing codes. To learn more about that, please read, “Motive, Means and Opportunity: Billing Codes Are Murdering Healthcare,” on 4sighthealth.com. Please appreciate the irony of organized medicine pushing hard for more billing codes to increase physicians’ revenue that someday may help put a few medical specialties out of business.
Anyway, here are the three medical specialties most vulnerable to substitution by AI and the share of payments they get for non-physical services:
- Pathology (99%)
- Diagnostic radiology (96%)
- Radiation oncology (62%)
The three medical specialties least vulnerable to substitution by AI and the share of payments they get for physical services:
- Dermatology (68%)
- Orthopedic surgery (53%)
- Ophthalmology (45%)
Interestingly, most of what primary care doctors do is a mix of physical and non-physical services, based on the researchers’ analysis. To wit, 90% of Medicare Part B payments to internal medicine specialists in 2023 were for a mix as was 87% of Medicare Part B payments to family medicine specialists. Maybe that mashup will make primary care less vulnerable to substitution by AI? Maybe not.
“The path for AI in different medical specialties depends not only on the technology, but on how it is designed, deployed, and reimbursed — whether payment and practice treat AI as a replacement or as a tool that makes work more attractive,” the researchers said.
In other words, we’ll pay for what we value more, humans or machines, not necessarily which one does a better job. Like The Who said, the simple things you see are all complicated.