Will AI Take Radiology? The 2026 Reality (Med School Premeds Read This)

In 2016, Geoffrey Hinton publicly told med students to stop training as radiologists. A decade later, the U.S. has a radiologist SHORTAGE — there are ~38,000 working radiologists and demand is growing 5% a year. AI did not replace them; it added a tool that scans every image first and flags the worrying ones, letting a single radiologist read 30-40% more studies per day. The lesson for any career: AI usually multiplies skilled workers before it replaces them.

Some examples

• The FDA had cleared 950+ medical AI devices by 2024 — most are radiology tools that augment, not replace (Aidoc, Viz.ai, RapidAI).
• Mayo Clinic's 2024 report shows radiologists using AI assist read 40% more studies per shift — but burnout dropped because boring scans got pre-triaged.
• Pathology, dermatology, and EKG reading have similar stories — AI accelerated the human, didn't unemploy them.
• Hinton himself walked back the comment in a 2024 60 Minutes interview, calling it 'one of the things I got wrong.'

Try it!

Pick a job you're considering. Search '[job title] AI augmentation 2025' on Google Scholar. Read the abstract of the first paper. You'll see the realistic version, not the LinkedIn-influencer hot take.