[antipaul]

If there was one application where deep learning was supposed to succeed, it was radiology

"people should stop training radiologists now" – Hinton, 2016

[Legend2440]

Per the article, it did succeed. AI radiology tools are being widely adopted, and they work very well.

But they are being used by radiologists, not instead of radiologists. And because scans can be interpreted more quickly and cheaply, more scans are ordered, which has increased the demand for radiologists overall.

[WhyOhWhyQ]

Probably a cliche, but https://en.wikipedia.org/wiki/Jevons_paradox