AI and radiologists: a painful divorce?

The statement made in November 2016 by Geoffrey Hinton, one of the godfathers of deep learning, that we should “stop training radiologists now” led to intense debate among radiologists whether artificial intelligence (AI) should be seen as an opportunity or a threat. Sessions on this topic are held at every radiology conference. The consensus seems to be that AI will not replace radiologists, instead, AI will give radiologists powerful tools that will help them to do a much better job. In March 2018, I was invited to talk in one such session with the title “AI and radiology: a perfect match”. To stir up some debate, I titled my talk: “AI and radiologists: a painful divorce?”. In the talk, I explain that AI will not only lead to powerful tools for radiologists but it will also automate certain tasks completely and it will provide image interpretation services to clinicians who then no longer need a radiology report to make treatment decisions.

As for Hinton’s statement, he repeated it in The New Yorker in 2017, but not in a recent JAMA viewpoint:

So no, we should definitively not stop training radiologists. But the profession of radiology will change profoundly as a result of the progress in AI. Don’t hold your breath: this profound change will take a decade or two. And it’s a good thing that will bring many improvements in healthcare, particularly in areas in the world where there are hardly any radiologists today.