I actually am trying to develop AI for medical imaging as an MRI physicist and well versed with medical imaging research (my day job) . At a MRI conference in Jan 2016 AI based scan we proposed by an eminent professor in the field. The community was frankly dimissive, fast forward 2019, 20% fields premier conference content is deep learning. The areas which are being developed area images synethsis (we go beyond traditional imaging physics) this gives new information via e.g. autoseq from Boston, scan speed up to get data more efficiently, near real interpretation, more importantly integration with other knowledge banks/graphs (e.g. genomics, vital health, electronic healh records). There are large players from tech e.g. Google, Facebook starting to add this to one of there businesses. NHS has recently launched a huge AI funding round. Medical imaging manufacturer already made huge investments to AI. University establishing AI research centres and doctoral training programs. And finally start up likes butterfly inc, which puts ultrasound scanner in a mobile phone, I still remember carry bulky pcb’s to fix ultrasound scanners, and I did like to think I’m not that old ;), this facts make re-think the role to radiologists. Where will and how they add value in 7–10 years time. This question is very different to role of a surgeon, robot have been around for a while to assist in (remote) operations but not to replace surgeona. Radiologists I’m not sure. Standard disclaimer a this personal opinion and not related to employer in any way.