Stop Training Radiologists: Deep Learning Wins Within Five Years
Geoffrey Hinton, 2016: 'People Should Stop Training Radiologists Now.' Five Years On, They Were Scarce
The deep learning pioneer compared radiologists to a cartoon coyote already over the cliff edge. By 2024 a radiology resident was writing about a historic shortage of radiologists.
What was said
“People should stop training radiologists now. It’s just completely obvious that within five years, deep learning is going to do a lot better than radiologists”
— Geoffrey Hinton (computer scientist, University of Toronto and Google) · said 2016 · due Within five years (by 2021)
Radiology kept training radiologists. Eight years on, a resident Hinton had told to reconsider wrote that the field faced 'a historic labor shortage'.
The fine print
Hinton said it on video in 2016; DOTmed reported the clip being shown to radiologists the following year. The full line opened: 'if you work as a radiologist, you're like the coyote that's already over the edge of the cliff, that hasn't yet looked down.' Another transcription has him adding: 'It might be 10 years, but we've got plenty of radiologists already.' Deep learning does beat radiologists on some narrow tasks; the forecast here is the one about the job.