eric topol. he's from the scripps research translational institute. dicine, how ai can make health care human again" and details how it can help with diagnosing decisions this is important topic. dr. topol, welcome back to "mad money." >> great to see you again. >> have a seat all right, doc, i read your book and i found it liberating because like many other people patients, i always feel like i'm being squeezed in. there's not enough time. ai gives doctors time to be doctors. >> exactly it's the ultimate gift of time by augmenting the performances of all clinicians and nurses that's a big decompression and to look at the data they're generating like a fly wheel you get the patient is taking on more. the doctors have to do less. the machine is coming to the rescue. >> can you give the viewers the situation -- i hate to -- cancer cancer is uniquely ai disease, right? >> yes well, it's because there are so many levels of data. there's -- not just the patient and their history, but the genome of the tumor. the scans that can be read through ai the slides,

joining us right now is the author of deep medicine, eric topol. good to see you.ead of the u.s. in terms of a.i. in healthcare. how, specifically? >> well, i spent the last year and-a-half as -- commissioned as a review for the government for the nhs. they started implementing a.i., getting rid of keyboards, so doctors don't do any typing, in emergency rooms, no less, patients love that because they have eye contact. reading scans, it revs up what a radiologist can do and the leader of the world in genomics. they're taking the data and they're implementing it into day-to-day practice. maria: that's exactly one of the jobs dagen that kai fu lee said would go away, radiologists. reading the radiology reading can be done quicker and easier by a machine, by a.i. dagen: how long will that be before that happens? maria: in the u.s. >> that's happening in china right now, more accurate than a lot of radiologists and of course a lot faster and a lot less expensive. but there is a concern that you want to have these algorithms at perfection level. if they're making a mista