i am pleased to have siddhartha mukherjee back at the stable. welcome. >> thank you, charlie. >> with some sense of fraternity, tell me how this piece came about. >> this piece came about because you know, it is a conference that you organized. >> rose: right. >> every fall, and i was talking at this conference to sebastian-- who works on ai. and sebastian told me, you know, i've been working on this thing in ai in medicine. and so my ears perked up. we were at the conference and my ears perked up. i said what are you doing? and sebastian than began to describe sort of an early version, this wasn't there yet. it certainly hadn't been published in nature as it eventually was. but it instantly caught my attention as a physician, what if we could use these powerful technologies, deep learning paradigms to start doing diagnosis. what would happen to radiology, what will happen to determine tolling. what will happen to doctors once we engage powerful computers to aid in diagnosis. so that is sort of the birth. >> that was-- so it began with a what it