Geoffrey Hinton, a leading AI researcher and Turing Award winner, now says he was too quick to declare that artificial intelligence would replace radiologists.
In a recent interview with the New York Times, Hinton admitted he focused too narrowly on image analysis and overestimated how quickly the field would evolve. Looking back, he says the general direction was right—but instead of replacing radiologists, AI is making them "a whole lot more efficient in addition to improving accuracy."
Back in 2016, Hinton famously claimed that training new radiologists was unnecessary, comparing the profession to a cartoon coyote running off a cliff without realizing there was no ground beneath. He argued that deep learning would outperform human radiologists within five years thanks to its broader experience. In a now widely shared video of the talk, fellow AI researcher and reinforcement learning expert Richard Sutton can be seen nodding in agreement.
AI helps with tasks instead of replacing professions
The reality has played out differently. The New York Times reports that institutions like the Mayo Clinic are showing how AI can help radiologists rather than replace them. Despite the rise of AI tools, the number of radiologists at Mayo Clinic has grown significantly—from around 260 in 2016 to over 400 today. That's a 55 percent increase.
Still, Hinton wasn’t entirely off base. AI is having a major impact on the profession. The clinic now uses more than 250 AI models in its radiology department, with some developed in-house and others sourced from external vendors. Many of these tools are also being used in cardiology.
The systems speed up image analysis, flag suspicious areas, and help detect conditions like blood clots or tumors. One model, for example, automatically measures kidney volume—a time-consuming task that once had to be done manually.
Dr. Matthew Callstrom, who leads the Mayo Clinic’s radiology department, describes AI as a second pair of eyes. It can take over repetitive tasks, but not the clinical judgment required in individual cases. According to Callstrom, AI-supported workflows are now part of everyday medical practice.
A lesson in humility for AI predictions
Hinton's earlier comments now serve as a cautionary tale. Many current claims—like those from OpenAI CEO Sam Altman—that AI will soon replace entire professions tend to oversimplify. They often fail to distinguish between automating or supporting individual tasks and eliminating whole job categories, much like Hinton did in 2016 when he reduced radiology to image analysis alone.
Even if large-scale automation were technically feasible, cultural, organizational, and legal factors are likely to slow down its rollout.
For AI researchers, there's a broader takeaway: avoid sweeping predictions about professions they may not fully understand. Not because being wrong is especially costly—few are ever held accountable—but because it shows respect for the people who carry out these jobs with real-world expertise and responsibility. If the medical field had actually followed Hinton’s advice in 2016 and stopped training radiologists, the consequences for patient care today could have been severe.
