Latest / kill switch / Algorithm, M.D.
Transcript
- 0:04Speaker 1: Sleepwalkers is a production of I Heart Radio and Unusual Productions.
- 0:12Speaker 1: Jan had become paralyzed through a disease process such that
- 0:17Speaker 1: she couldn't move anything below her neck, and she volunteered
- 0:21Speaker 1: to have the surgical procedure in which we implanted electrodes
- 0:25Speaker 1: in her head and we were able to decode the
- 0:29Speaker 1: signals from her brain and she was able to move
- 0:31Speaker 1: a high performance prosthetic arm in hand. That's Andy Schwartz,
- 0:37Speaker 1: a professor of neurobiology at the University of Pittsburgh. He
- 0:41Speaker 1: helped build the technology that allowed a patient called Jan
- 0:43Speaker 1: Sherman to control a robotic arm with her mind. In
- 0:48Speaker 1: order to do that, Andy's team needed to give the
- 0:50Speaker 1: arm away to know where Jan wanted it to move.
- 0:53Speaker 1: They had to read signals directly from her brain and
- 0:56Speaker 1: teach a computer to understand them. So they built a
- 0:59Speaker 1: brain computer interface a b c I. We implanted electrodes
- 1:03Speaker 1: in her head. You have to put on these bulky
- 1:05Speaker 1: connectors with thick cables going to a bank of amplifiers
- 1:08Speaker 1: and computers. It sounds like something out of a science
- 1:11Speaker 1: fiction film. But once Andy could see Jan's neurons firing,
- 1:15Speaker 1: reading her intentions was simpler than you might think. What
- 1:20Speaker 1: we found is that the rate that these neurons fire
- 1:24Speaker 1: is related to the direction that the arm moves. When
- 1:27Speaker 1: you add their signals together, you can get a very
- 1:29Speaker 1: precise representation of that movement. It's a very very simple algorithm,
- 1:34Speaker 1: and it's like listening to a Geiger counter um and
- 1:38Speaker 1: each click of the Geiger counters the same, but as
- 1:41Speaker 1: you get closer to a radioactive source, those clicks come
- 1:45Speaker 1: closer together. Listening for those clicks was the breakthrough necessary
- 1:50Speaker 1: to translate Jan's thoughts into signals that the robotic arm
- 1:53Speaker 1: could understand, and this meant that Jan could move again.
- 1:57Speaker 1: She could reach out and touch her husband, and Andy's
- 2:01Speaker 1: team filmed Jan using the arm to complete a more
- 2:04Speaker 1: playful goal that she'd set to feed herself chocolate for
- 2:08Speaker 1: the first time in ten years for a woman, one
- 2:16Speaker 1: giant fight for DHDR. Jan was able to do that
- 2:22Speaker 1: with the robot, and not only that, but she made
- 2:25Speaker 1: graceful and beautiful movements. That's what really blew me away.
- 2:29Speaker 1: It looks much like a real arm in hand, and
- 2:33Speaker 1: for someone who studies movement, it was really quite beautiful.
- 2:40Speaker 1: Andy is a scientist and a researcher not given to
- 2:43Speaker 1: being sentimental, but the way he talks about Jan moving
- 2:46Speaker 1: her robotic arm, it's like describing a dance a ballet.
- 2:50Speaker 1: It's one of the most inspiring examples of humans and
- 2:53Speaker 1: machines working together in concert that we've come across in
- 2:56Speaker 1: all of our reporting for Sleepwalkers. But it also raises
- 3:00Speaker 1: profound questions about the future of our health, our bodies,
- 3:04Speaker 1: and our society. What are the implications positive and negative?
- 3:08Speaker 1: As AI makes us ever more able to decode complex
- 3:12Speaker 1: systems like the brain or the human genome, how is
- 3:16Speaker 1: AI poised to change the world of medicine. I'm as
- 3:20Speaker 1: Veloshin and this is Sleepwalkers, So Kara I found Andy's
- 3:36Speaker 1: story completely mind blowing, no pun intended. You know it
- 3:41Speaker 1: shakes one of the last remaining mysteries. For the most part,
- 3:44Speaker 1: neuroscience is still very much a black box problem. We
- 3:47Speaker 1: know what's happening in the brain, but neuroscientists can't always
- 3:51Speaker 1: know why and how, which is a lot like the
- 3:54Speaker 1: problem of black box AI. We are moving quickly into
- 3:57Speaker 1: a world where sensors can read us better and better
- 4:00Speaker 1: know they can read pupil dilation, carbon dioxide in the
- 4:03Speaker 1: breath to understand people's emotions. Yeah, this is the stuff
- 4:06Speaker 1: we talked about an episode four around using AI to
- 4:09Speaker 1: better read biometrics with Poppy Chrum. The difference being in
- 4:12Speaker 1: this case is that Andy isn't monitoring the outside of
- 4:14Speaker 1: our bodies, so the privacy concern is less. To read
- 4:18Speaker 1: Jan's brain, he had to drill into her skull and
- 4:21Speaker 1: place electros into the surface of her brain and then
- 4:25Speaker 1: connect them to a computer, so that's unlikely to creep
- 4:28Speaker 1: up on you. Google is not going to be doing
- 4:30Speaker 1: that to get geolocation, not yet. Because that said, and
- 4:35Speaker 1: he told me one of his big priorities is trying
- 4:37Speaker 1: to figure out how to achieve the same effects without
- 4:40Speaker 1: needing invasive surgery, so that people can use this technology
- 4:43Speaker 1: at home. Talking about reading the brain, one cutting edge
- 4:46Speaker 1: application for AI is to restore language. A few months ago,
- 4:50Speaker 1: I was actually reading this paper in Nature, and I'm
- 4:52Speaker 1: not sure all podcast listeners read Nature. Well, thanks, Carol,
- 4:56Speaker 1: you do the hard work so that we don't have to.
- 4:58Speaker 1: I guess why might they pay me the big bitcoin um?
- 5:01Speaker 1: But you know, basically this article was about decoding the
- 5:03Speaker 1: human brain to produce language based on how the brain
- 5:07Speaker 1: tells the mouth to move well. Funny enough, this is
- 5:11Speaker 1: something I spoke to Andy about a few months before
- 5:15Speaker 1: your paper and Nature was published, Kara, and he was
- 5:18Speaker 1: talking about exactly this. So if you record from motor
- 5:21Speaker 1: areas associated with producing language, you can start to recognize
- 5:26Speaker 1: certain words and phrases. Even I think that's realizable in
- 5:30Speaker 1: the your term. What Andy is saying is that to
- 5:33Speaker 1: create spoken words directly from the brain, we don't actually
- 5:36Speaker 1: need to read thoughts. We can just look at the
- 5:38Speaker 1: last step, the moment when thought becomes action as neurons
- 5:42Speaker 1: fired to move our tongue, our lips are jaw to
- 5:45Speaker 1: create sound. And just like with jan, an algorithm can
- 5:49Speaker 1: listen to those neurons and allow thoughts to become actions.
- 5:53Speaker 1: And this could transform lives. So if you could start
- 5:57Speaker 1: to recognize words and language from brain activity, that would
- 6:02Speaker 1: be very helpful for people who are locked in with
- 6:05Speaker 1: a l S and can't communicate. We talked in the
- 6:08Speaker 1: last episode about how many technological breakthroughs have come out
- 6:12Speaker 1: of DARPER. That's the branch of the Defense Department charged
- 6:15Speaker 1: with inventing technological surprises, well neural prosthetics or robotic limbs
- 6:20Speaker 1: controlled by the mind have been an area of heavy
- 6:23Speaker 1: investment for the agency. You may remember Gil Pratt from
- 6:26Speaker 1: earlier in the series. He's now the CEO of the
- 6:29Speaker 1: Toyota Research Institute, but previously he was at DARPA where
- 6:33Speaker 1: he worked with none other than Andy Schwartz. So he
- 6:37Speaker 1: was involved in a project at DARPA which was called
- 6:39Speaker 1: Revolutionizing Prosthetics. And the project that I started was to
- 6:45Speaker 1: see if we could actually help some of the experimental
- 6:48Speaker 1: patients that he had to perform even better. So Kara
- 6:52Speaker 1: I was interviewing Gil Pratt about his work at Toyota
- 6:55Speaker 1: and on self driving cars, and we got talking about
- 6:58Speaker 1: his interest in these human machine owner ships, and I said,
- 7:01Speaker 1: let me tell you a story as a scientist called
- 7:04Speaker 1: Andy Schwartz and his patient Jan and Gil was like, um, yeah,
- 7:08Speaker 1: I worked on that. Yeah, no, no no, no, literally, it's funny.
- 7:13Speaker 1: And it also shows the long arm of DARPA intended.
- 7:17Speaker 1: Again that's far away from military applications. You know, darp
- 7:21Speaker 1: AS interest in prosthetics is actually because of veterans, many
- 7:25Speaker 1: of whom lose limbs on the battlefield, which highlights again
- 7:29Speaker 1: the dual use nature of so much innovation. Right, um
- 7:32Speaker 1: and The work Gil was doing at DARPA with Jan
- 7:35Speaker 1: was all about doing about a job of interpreting her
- 7:37Speaker 1: brain waves using the existing models. My group made a
- 7:42Speaker 1: system called arm assist that watched what Jan was trying
- 7:46Speaker 1: to do. It inferred Okay, now she's trying to pick
- 7:49Speaker 1: up a block. Now she's trying to move it over
- 7:51Speaker 1: to the left. Now she's trying to drop the block.
- 7:54Speaker 1: You know, in a way very similar to if you
- 7:55Speaker 1: use power Point and you have like snapped to grit
- 7:59Speaker 1: or snapped the object turned on, it will help you
- 8:02Speaker 1: move the mouse to where it thinks you want to go.
- 8:05Speaker 1: This system helped her move the arm to where it
- 8:08Speaker 1: inferred she wanted to go based on a very noisy
- 8:11Speaker 1: signal that was coming out of her brain. That noise
- 8:13Speaker 1: in the signal was partly because the connection between Jan's
- 8:16Speaker 1: brain and the decoding computer was weak. So Gil's team
- 8:20Speaker 1: Adoppa designed a program to boost Jan's intentions. We tested
- 8:25Speaker 1: her by randomly turning the assists on and off. And
- 8:29Speaker 1: you can think of the assist as a guardian, like
- 8:32Speaker 1: what we're developing for the car to stop you from
- 8:34Speaker 1: having a crash, and we had the guardian in this
- 8:38Speaker 1: case help as little as possible, but still be effective
- 8:42Speaker 1: at helping her to reach the goal. The amount of
- 8:45Speaker 1: help that we gave her was so low that she
- 8:48Speaker 1: couldn't tell whether it was on or off. But when
- 8:51Speaker 1: Guardian was turned off, Jan's success rate fell. And in
- 8:55Speaker 1: a human machine partnership is not just the computer that
- 8:58Speaker 1: adapts to the human brain also adapts to the algorithm.
- 9:03Speaker 1: And the's algorithms are simple and they rely on human
- 9:06Speaker 1: programmed rules. So when they see cause in this case,
- 9:10Speaker 1: enough neurons firing at the same time, they create an effect,
- 9:14Speaker 1: a movement. But what's tantalizing is that effect that output
- 9:19Speaker 1: hints at something far deeper and infinitely complex personality. I
- 9:25Speaker 1: like Moby Dick where he talks about Captain Ahab walking
- 9:29Speaker 1: on the deck of the ship, and by observing his movements,
- 9:31Speaker 1: you can really understand what he's thinking about. If you
- 9:34Speaker 1: think about movement, it's really a communication between your innermost
- 9:39Speaker 1: thoughts in the outside world. Andy and his team saw
- 9:42Speaker 1: this come to life before their very eyes when they
- 9:45Speaker 1: connected another subject, Nathan, to a neural prosthesis. The robotic
- 9:50Speaker 1: arm moved in accordance with the personality of the user
- 9:54Speaker 1: Jan when she moved. She was very careful and gentle,
- 9:58Speaker 1: and Nathan is more are of a video gamer. He's
- 10:01Speaker 1: a younger guy. He's more of a competitor, so he
- 10:03Speaker 1: would move much faster, and so he would pick up
- 10:07Speaker 1: an object and then instead of placing it carefully in
- 10:10Speaker 1: a receptacle, he would basically toss it into the receptacle
- 10:13Speaker 1: to be faster. Today we can see the difference in
- 10:17Speaker 1: Nathan and Gen's personalities from how the brain controls a
- 10:20Speaker 1: prosthetic arm. We can infer personality from output, but we
- 10:25Speaker 1: don't even have sophisticated enough tools to ask why. As
- 10:29Speaker 1: we get better and better at these computational approaches, will
- 10:32Speaker 1: gain a much better understanding of the way the brain works.
- 10:37Speaker 1: Rather than having some major event causing some simple consequence. Instead,
- 10:43Speaker 1: it's more like a perfect storm where many factors come
- 10:46Speaker 1: together to generate a consequence. And if we understand which
- 10:52Speaker 1: of those events are important and how they're combined, then
- 10:56Speaker 1: we can start to understand brain function better. If you
- 11:01Speaker 1: see a cup and you're thirsty, if we could realize
- 11:04Speaker 1: that what you want to do is to drink from it,
- 11:07Speaker 1: then we could understand that you want to grasp the
- 11:10Speaker 1: cup from the side, and if we could distinguish that
- 11:14Speaker 1: from you grasping a cup with the intention of passing
- 11:17Speaker 1: it to me, you might hold it now from the top,
- 11:20Speaker 1: then we could do a better job generating the correct
- 11:23Speaker 1: movement for Andy. The next frontier is more deeply understanding
- 11:27Speaker 1: the human brain carrot by developing new models and better algorithms.
- 11:31Speaker 1: And that's really a computer science problem as much as
- 11:34Speaker 1: a medical problem. And then really the next frontier in
- 11:36Speaker 1: medicine is all about using AI to decode complex interactions.
- 11:41Speaker 1: And in fact, you reported a piece on exactly this phenomenon. Yeah,
- 11:45Speaker 1: I spoke to a woman named Regina bars Ali. She's
- 11:48Speaker 1: a computer scientist m I T and her own experience
- 11:51Speaker 1: of how she was diagnosed with breast cancer actually inspired
- 11:54Speaker 1: her to work on bringing AI into the realm of medicine.
- 11:58Speaker 1: When we come back, we'll here from Regina and take
- 12:01Speaker 1: a look at other ways AI is changing diagnostics and
- 12:04Speaker 1: the future of our health. I remember still, I went
- 12:15Speaker 1: to mamogram and they told me a high density but
- 12:18Speaker 1: you shouldn't worry. Half of the population have high density breasts,
- 12:22Speaker 1: so don't worry about it. That's Regina bars Lie. Regina
- 12:26Speaker 1: is a professor of Electrical Engineering and Computer Science at
- 12:29Speaker 1: m I T. I couldn't believe in it. I didn't
- 12:31Speaker 1: feel anything. There was nothing wrong, you know. I was
- 12:34Speaker 1: continuing my morning runs and being fine. It was later
- 12:37Speaker 1: on after her mammogram that Regina found out she had
- 12:40Speaker 1: breast cancer, and as it often does, the news came
- 12:44Speaker 1: as a surprise. And if I'm looking at myself, I
- 12:47Speaker 1: really cannot explain what was wrong that I got this
- 12:51Speaker 1: disease that clearly didn't have any family history. I'm exercising,
- 12:55Speaker 1: I'm meeting healthy and for many many patients that I'm
- 13:00Speaker 1: me during my own journey, UH, their diagnosis came to
- 13:03Speaker 1: them as the biggest surprise. The thing is, according to
- 13:07Speaker 1: current medical standards, breast cancer risk is based on a
- 13:10Speaker 1: few factors. Are you a woman, and are you old?
- 13:13Speaker 1: Do you have the Brocko gene? And do you have
- 13:14Speaker 1: breast cancer in your family? But those are relatively simple
- 13:18Speaker 1: inputs and they don't account for what complex systems we
- 13:21Speaker 1: are above eighty percent of women who diagnosed with breast
- 13:25Speaker 1: cancer there first in their families, so it's not clear
- 13:30Speaker 1: you know what causes it. According to the Susan Becoman
- 13:33Speaker 1: Breast Cancer Foundation, breast cancer is the most common cancer
- 13:37Speaker 1: amongst women around the world. Every two minutes a case
- 13:40Speaker 1: of breast cancer is diagnosed and a woman in the
- 13:42Speaker 1: United States, and every minute somewhere on earth a woman
- 13:46Speaker 1: dies of breast cancer. That's more than four hundred women
- 13:50Speaker 1: per day. So if you're listening to this podcast, you
- 13:53Speaker 1: probably know someone who has been or will be diagnosed
- 13:56Speaker 1: with breast cancer in their lifetime. And with that many
- 13:59Speaker 1: people affected that they are bound to be some oversights,
- 14:02Speaker 1: like in Regina's case. I discovered that my own diagnosis
- 14:06Speaker 1: was delayed because the malignancy was missed in the previous mammograms.
- 14:10Speaker 1: And I also discovered that this is not a unique experience.
- 14:15Speaker 1: So the question I asked is is it possible for
- 14:18Speaker 1: us to know ahead of time what's to come? In
- 14:22Speaker 1: other words, can we look into the future prior to
- 14:26Speaker 1: our diagnosis. Regina had already been a long time computer
- 14:29Speaker 1: scientists at M I T. She thought often about machine
- 14:32Speaker 1: learning in her work, but this new personal hardship redirected
- 14:35Speaker 1: her thinking. When you go to the hospital, you see
- 14:39Speaker 1: like real pain of other people. You see people who
- 14:42Speaker 1: go through key most radiation, even though the hospital just
- 14:46Speaker 1: want stop away from M I T. I just was
- 14:49Speaker 1: not aware there is so much suffering. And at that
- 14:52Speaker 1: point when I came back, I was thinking, we create
- 14:56Speaker 1: so much you know, exciting new technology it and I
- 14:59Speaker 1: see why I we are not trying to solve it.
- 15:03Speaker 1: It's it's just a travesty. And so Regine and her
- 15:07Speaker 1: colleagues that M I T began training a deep learning
- 15:09Speaker 1: model on over ninety thod mamograms from mass General Hospital,
- 15:14Speaker 1: and with such a large data set, they were able
- 15:16Speaker 1: to predict a patient's risk of breast cancer by comparing
- 15:19Speaker 1: one mammogram to tens of thousands of others instantly. My
- 15:23Speaker 1: firm belief was that despite the standard risk factors, there
- 15:27Speaker 1: is a lot of information in women's breast. Human eye
- 15:31Speaker 1: which even seen you know, thousands or tens of thousands
- 15:35Speaker 1: images over their lifetime, may not be really able to
- 15:39Speaker 1: detect it with a great clarity. However, if machine which
- 15:43Speaker 1: is uh you know, trained on sixty images where it
- 15:46Speaker 1: knows the outcomes, it can identify the difference in pixel
- 15:51Speaker 1: distribution that I likely correlate to a future things that
- 15:57Speaker 1: may come. The amount of data which you can train
- 16:00Speaker 1: a computer on versus a doctor is massive and the
- 16:04Speaker 1: AI was able to detect smaller details than the human
- 16:07Speaker 1: I could pick up, so this does feel like a
- 16:09Speaker 1: perfect application for the strengths of machine learning. Fifteen months
- 16:15Speaker 1: ago we put first our density model, which is every
- 16:19Speaker 1: mammograms that goes through months general it shows a prediction
- 16:22Speaker 1: to the radiologists and around of the times radiologists degree
- 16:27Speaker 1: with the machine, and when they disagree, the more experienced
- 16:31Speaker 1: radiologists typically sides up with the machine. So Regina's early
- 16:35Speaker 1: efforts are doing really well, and the hope is that
- 16:38Speaker 1: more hospitals around the country will begin using these models
- 16:41Speaker 1: for early detection and risk assessment. And that's not only
- 16:43Speaker 1: because understanding risk is important, but also because misdiagnoses have
- 16:49Speaker 1: led to unnecessary surgeries. I read this book The Emperor
- 16:53Speaker 1: Whole Melodies. It's a book about comes, so I really
- 16:56Speaker 1: like let me just say one particular moment it really
- 17:00Speaker 1: choices me out. It's about how men, male surgeons who
- 17:04Speaker 1: are treating women with the surgeries this firmly believe that
- 17:09Speaker 1: the more you cut out of you know, women's body,
- 17:12Speaker 1: the beta is w likelihood. In the early twentieth century,
- 17:16Speaker 1: there was an insurgence of doctors performing radical mastectomies, removing
- 17:20Speaker 1: the entire breast, thus permanently disfiguring a patient, and radical
- 17:25Speaker 1: mastectomies became the norm for much of the twentieth century.
- 17:28Speaker 1: The reasoning was that removing a lump leaves the risk
- 17:31Speaker 1: of tumors growing elsewhere, so why take the risk. By
- 17:35Speaker 1: the nineteen eighties, it was clear that radical mastectomies weren't
- 17:38Speaker 1: actually in effective treatment for many patients, but according to Regina,
- 17:42Speaker 1: surgeons still remove too much tissue out of an abundance
- 17:45Speaker 1: of caution. The reason it happens, it's not because there
- 17:48Speaker 1: is some evil doctors that wants to look another surgery.
- 17:51Speaker 1: It's just because people are uncertain and it's high risk,
- 17:54Speaker 1: and many times conceptician would say, I am ready to
- 17:56Speaker 1: go for the harshest treatment to minimize that chances. So
- 18:01Speaker 1: what we demonstrated that with machine learning you can actually
- 18:05Speaker 1: identify a city percent of the populations it doesn't need
- 18:09Speaker 1: this type of surgery. Regina told me that had these
- 18:13Speaker 1: deep learning models been in place at the time of
- 18:15Speaker 1: her early mamograms, she might have detected her risk two
- 18:18Speaker 1: years sooner. And in many cases, early detection makes a
- 18:22Speaker 1: big difference in how a patient chooses to treat their cancer.
- 18:25Speaker 1: Since developing the breast cancer detection model, Regina is now
- 18:28Speaker 1: co leading m I T S J Clinic, which is
- 18:31Speaker 1: a new initiative focusing on machine learning and health. What
- 18:34Speaker 1: I hope that we as a society advance since then
- 18:39Speaker 1: and we are ready to bring you know, the recent
- 18:42Speaker 1: science and help women, and even if it means that
- 18:46Speaker 1: we need to change our beliefs about how risk assessment works.
- 18:52Speaker 1: And Regina hopes that as a society we can move
- 18:54Speaker 1: toward greater acceptance of using machine learning to enhance medicine.
- 18:58Speaker 1: Whichever one does a bit, Joel, that one should prevail.
- 19:05Speaker 1: So big question, how do you feel, Kara about putting
- 19:08Speaker 1: your health into the hands of an algorithm? You know,
- 19:11Speaker 1: after speaking to Regina, who told me that she probably
- 19:15Speaker 1: would have been diagnosed two years sooner using her models,
- 19:20Speaker 1: it seems as though machine learning provides this unparalleled form
- 19:23Speaker 1: of detection right because the most seasoned doctors simply can't
- 19:28Speaker 1: compare thousands of data points at once. I don't think
- 19:31Speaker 1: the issue is algorithms replacing doctors. It's more a matter
- 19:34Speaker 1: of equipping doctors the sharper tools that they can do
- 19:37Speaker 1: their jobs. They've got to provide a patient with information
- 19:40Speaker 1: and allow that patient to make informed decisions. Algorithms don't
- 19:44Speaker 1: have a bedside manner. You know. What you say about
- 19:46Speaker 1: the shop of tools reminds me of our conversation about
- 19:48Speaker 1: creativity and episode two, using AI to give artists, musicians,
- 19:53Speaker 1: screenwriters new tools to do better work. At the same time,
- 19:58Speaker 1: just like the art world, the medical profession sits on
- 20:00Speaker 1: this enormous pedestal where we have to trust what they
- 20:03Speaker 1: say because most of us don't have the tools to
- 20:06Speaker 1: question them, right unless you're on WebMD. Yeah, the bane
- 20:09Speaker 1: of every doctor's life. Yeah, you know, doctors have a
- 20:11Speaker 1: hard enough time explaining medicine to patients. Imagine them having
- 20:14Speaker 1: to explain artificial intelligence. Well, from experience, we can say
- 20:17Speaker 1: good luck to him. What's crazy to me actually is
- 20:21Speaker 1: that Regina and her co author, this woman, doctor Connie Lehman,
- 20:24Speaker 1: who's a radiologist at Mass General, were rejected from every
- 20:28Speaker 1: single federal grant they submitted at first. Why would they
- 20:30Speaker 1: be rejected from all those federal grants? Because as much
- 20:33Speaker 1: as there is a ton of buzz surrounding AI, I
- 20:36Speaker 1: think people have to appreciate how new this frontier is right,
- 20:40Speaker 1: and using machine learning to make predictions about people's cancer
- 20:44Speaker 1: is very, very new, and it's going to take doctors
- 20:47Speaker 1: a really long time to learn how to convey this
- 20:49Speaker 1: information to patients. So that's where Regina is right now,
- 20:53Speaker 1: figuring out how doctors can explain to patients, Hey, AI
- 20:57Speaker 1: helped us determine your cancer risk. Well, part of problem
- 21:00Speaker 1: is that we don't yet have explainable AI. So it's
- 21:03Speaker 1: not just that it's hard to explain it to patients,
- 21:05Speaker 1: it's actually a black box. You may remember Sebastian Thrun,
- 21:09Speaker 1: the founder of Google X from earlier in the series.
- 21:12Speaker 1: As well as self driving cars and flying cars. He
- 21:15Speaker 1: works on medical diagnostics and he recognizes a problem. One
- 21:20Speaker 1: of the conundrums of machine learning is that when you
- 21:23Speaker 1: open up in doing that brook you look at like
- 21:25Speaker 1: hundreds of millions of numbers, but he can't quite understand
- 21:28Speaker 1: what's happening. So people are are concerned. People look at
- 21:31Speaker 1: that Brooks say, add the thing of diagnals and cancer,
- 21:33Speaker 1: what does it do. We've talked about the black box
- 21:36Speaker 1: problem in AI and how hard it is to trust
- 21:39Speaker 1: decisions that can't be explained, but Sebastian is quick to
- 21:42Speaker 1: point out that human beings can also be difficult to decipher.
- 21:47Speaker 1: Let's remind ourselves our doctors are also black boxes. You
- 21:50Speaker 1: can't open up the brain of your doctor and ask
- 21:53Speaker 1: what was this here she using for diagonals and cancer.
- 21:56Speaker 1: It's a fair point, and it's one has also been
- 21:58Speaker 1: noted by Siddathan Kache. He is one of the world's
- 22:02Speaker 1: foremost cancer doctors and the Pulitzerprise winning author of The
- 22:05Speaker 1: Emperor of All Maladies. That's the book that Regina referred
- 22:09Speaker 1: to earlier, and Siddartha has also written extensively about how
- 22:13Speaker 1: AI is changing medicine. I've been a huge fan of
- 22:16Speaker 1: his work for a long time and actually ambushed him
- 22:18Speaker 1: after a talk he gave in order to persuade him
- 22:21Speaker 1: to an interview for this podcast. So thinking about AI
- 22:24Speaker 1: helping diagnose patients, it's worth asking is the human doctor
- 22:28Speaker 1: so very different? One problem that I think is fascinating
- 22:32Speaker 1: is when a patient comes into the hospital. If you
- 22:35Speaker 1: ask a particularly astute physician, that physician can actually describe
- 22:39Speaker 1: to you what the most likely journey of that patient
- 22:42Speaker 1: will be in the hospital, whether they're likely to stay
- 22:44Speaker 1: for twenty five days suffered through bacterial stepsis, you know
- 22:47Speaker 1: all from peeking in through the door of in emriency room,
- 22:50Speaker 1: Siddartha started to wonder how doctors make those lightning fast calls,
- 22:54Speaker 1: and it got him interested in understanding what the brain
- 22:57Speaker 1: is doing when a doctor makes a die ignosis. We
- 23:01Speaker 1: actually understand very little about how human beings make diagnoses.
- 23:05Speaker 1: I mean, the studies that have been done so far
- 23:07Speaker 1: suggest that most people make diagnosis in a kind of
- 23:11Speaker 1: recognition sense rather than an algorithmic sense. The classical description
- 23:15Speaker 1: of how we make diagnosis was extraordinarily algorithmic, sort of
- 23:18Speaker 1: goes down a series of elimination. It's not this, it's
- 23:22Speaker 1: not that. Now. Whence the alto says algorithmic, he doesn't
- 23:25Speaker 1: mean using a computer algorithm. He means using rule based logic.
- 23:29Speaker 1: If this, then that, et cetera. But what he learned
- 23:33Speaker 1: was that despite what the textbooks say, that's not actually
- 23:36Speaker 1: how doctors make a diagnosis. When you put doctors inside
- 23:40Speaker 1: MRI machines and ask the question how do they make diagnosis?
- 23:44Speaker 1: In fact what lights up is parts of the brain
- 23:47Speaker 1: that are much much more to do with pattern recognition.
- 23:50Speaker 1: Here's a rhinoceros, Here's not a rhinoceros. Here's an elephant,
- 23:53Speaker 1: here's not an elephant. Especially mature doctors make diagnoses based
- 23:58Speaker 1: on patent recognition, and they'll flip around like moths around
- 24:02Speaker 1: the flame and ultimately slowly arrive at the target. It's
- 24:05Speaker 1: much more geographical way of thinking rather than linear. They're
- 24:09Speaker 1: using a combination of Bayesian or prior probability understandings. They're
- 24:13Speaker 1: using pattern recognition. They're understanding things about the patient and
- 24:17Speaker 1: figuring out what to do. Hearing Saddatha speak about doctor's
- 24:24Speaker 1: cara in terms like prior probability understandings and Bayesian statistics
- 24:29Speaker 1: really does make it sound like he's describing AI rather
- 24:32Speaker 1: than people. Well, it kind of is. You know, neural
- 24:34Speaker 1: networks are purposely modeled on the human brain. It's not
- 24:38Speaker 1: as easy as causing effect. It's about drawing on a
- 24:41Speaker 1: lifetime of experience to make best guesses based on competing
- 24:43Speaker 1: information that we have to weigh appropriately in micro second.
- 24:47Speaker 1: It's no easy task. It's funny because we've worn several
- 24:50Speaker 1: times on this series that we shouldn't be surprised when
- 24:52Speaker 1: our creations reflect us, and yet it's almost impossible not
- 24:56Speaker 1: to be. I feel a sense of uncanny chills when
- 24:59Speaker 1: Saddatha is gribes a human doctor working like an algorithm.
- 25:03Speaker 1: And he wrote about this in The New Yorker with
- 25:05Speaker 1: the headline AI versus m D And he made this point,
- 25:09Speaker 1: which is that human and machine processes of making diagnosis
- 25:12Speaker 1: are converging. And it makes me wonder who's going to
- 25:17Speaker 1: have the final word. Well, I asked Sebastian thrun exactly
- 25:21Speaker 1: that question. He will die of cancer a lot. I
- 25:24Speaker 1: believe many of those deaths is actually preventable using artificial intelligence.
- 25:29Speaker 1: It's amazing how diverse diagnostics you get. When you show
- 25:32Speaker 1: a set of dermatologists the same set of images, some
- 25:35Speaker 1: will say cancers as I would say five. And Sebastian
- 25:40Speaker 1: has a personal interest in the topic. My family, unfortunately
- 25:44Speaker 1: has a long, long, long history of cancer. My my
- 25:47Speaker 1: sister passed away last year. My mother passed away a
- 25:50Speaker 1: young age. So one of my questions I had in
- 25:52Speaker 1: my life with me is since my mother died, um,
- 25:56Speaker 1: maybe we should not work on on treatment. We should
- 25:58Speaker 1: really focus on detect not diagnostics. Diagnosis of skin cancer
- 26:03Speaker 1: doesn't require looking inside your organs. You can just look
- 26:05Speaker 1: at the person from outside and it turns out we're
- 26:08Speaker 1: not heavenal symptoms. Before it becomes dangerous, it sits therefore
- 26:11Speaker 1: quite a vie. It grows below your skin, it spreads,
- 26:14Speaker 1: and then it destroys your liver, and then your first
- 26:16Speaker 1: symptom might be that back pain or a yellow face.
- 26:19Speaker 1: Maybe we should just look every single day. In fact,
- 26:23Speaker 1: Sebastian his work to make it possible for people to
- 26:25Speaker 1: check themselves every day. He published a paper in Nature
- 26:29Speaker 1: called Dermatologist level classification of skin cancer with Deep neural Networks.
- 26:35Speaker 1: What he demonstrated is that a program that runs on
- 26:38Speaker 1: an iPhone performs just as well as dermatologists at diagnosing
- 26:42Speaker 1: skin cancer. It sounds transformative, but Siddhartha has a very
- 26:47Speaker 1: specific concern about this kind of technology entering the mainstream.
- 26:51Speaker 1: Well Over diagnosis is an important risk. A classic example
- 26:54Speaker 1: of that is a lesion in the breast, a spot
- 26:58Speaker 1: that is actually not reast cancer, but it's picked up
- 27:01Speaker 1: and described as breast cancer, that leads to a biopsy,
- 27:04Speaker 1: the barbsy leads to complications and so forth, and at
- 27:07Speaker 1: the end of it you discover that that you know
- 27:08Speaker 1: you've achieved not very much, except for subjecting a woman
- 27:11Speaker 1: to an unpleasant procedure with unpleasant costs. We don't want
- 27:15Speaker 1: to catch just early cancer. We want to catch the
- 27:17Speaker 1: early cancers that are likely to kill you, the other
- 27:20Speaker 1: ones that are unlikely to become anything. We actually want
- 27:24Speaker 1: to be able to reassure patients that they don't need
- 27:27Speaker 1: a biopsy. Regina told Kara how screening enabled by machine
- 27:31Speaker 1: learning could reduce unnecessary mystectomies, But according to Siddartha, we
- 27:36Speaker 1: also need to be cautious about overscreening pushing us into
- 27:39Speaker 1: unnecessary procedures. And as AI and sensors become more ubiquitous,
- 27:44Speaker 1: enabling us to constantly search for illness, there may be
- 27:48Speaker 1: psychological implications that were not fully prepared for. This is
- 27:52Speaker 1: the very or valiant notion of previvors. It's the word
- 27:55Speaker 1: that I first encountering clinic in it was a woman
- 27:57Speaker 1: who had brack of one mutation, but in fact did
- 28:00Speaker 1: not have any breast cancer. She called herself a pre
- 28:02Speaker 1: vivor of breast cancer. She was a survivor of a
- 28:05Speaker 1: disease that she yet did not have. Our culture hasn't
- 28:08Speaker 1: reached the place that you know, we're routinely thinking of
- 28:10Speaker 1: ourselves as previvors. But it has reached a place where
- 28:13Speaker 1: surveillance is is constant. You know, you're moving from colonoscopy
- 28:16Speaker 1: to mammogram to p s A test, to medical exam
- 28:21Speaker 1: to retinal exam. And you can imagine stringing together with
- 28:26Speaker 1: future devices, a culture in which the body is always
- 28:29Speaker 1: being hunted scoured for being a potential locus of future disease,
- 28:33Speaker 1: and that will I think distrought culture fundamentally. It's a
- 28:36Speaker 1: very Orwell and very scary idea, said Arthur Ludes. Of course,
- 28:40Speaker 1: Sir George Orwell, whose novel four was prescient about the
- 28:44Speaker 1: culture of surveillance that's now blooming around us. But I
- 28:47Speaker 1: never thought about surveillance in medical terms before. And who
- 28:51Speaker 1: might be surveilling our bodies. One might be a health
- 28:54Speaker 1: insurance company or the government interested in, you know, who's
- 28:58Speaker 1: healthy and who's not healthy. There was a chance meeting
- 29:01Speaker 1: between Siddartha and Sebastian that God Siddartha thinking about AI
- 29:05Speaker 1: and medicine, but the two have fundamental disagreements on the
- 29:08Speaker 1: risks and rewards of surveiling the body. I love Sid
- 29:11Speaker 1: as a person, but I can tell you any doctor
- 29:14Speaker 1: who tells you less data is better for you is irresponsible.
- 29:18Speaker 1: If I could give you information with your skin cancer
- 29:21Speaker 1: every day, you will live longer than if it's just
- 29:24Speaker 1: consulted avntologies every year or two. But also the unpredictability
- 29:29Speaker 1: of death is part of the human experience. Our culture
- 29:32Speaker 1: would be very different if we walked around with signs
- 29:34Speaker 1: on our foreheads which told us the number of days
- 29:37Speaker 1: that we had left to live. What Siddartha is describing
- 29:40Speaker 1: is not some thought experiment. Using AI to predict time
- 29:44Speaker 1: of death is fast becoming a reality. But what inputs
- 29:48Speaker 1: does it use? And how am I knowing when we
- 29:50Speaker 1: will die? Change our culture? Join us after the break.
- 30:02Speaker 1: Doctors are actually very poor at predicting death. If you
- 30:05Speaker 1: look at the pattern of how people die, most people
- 30:09Speaker 1: don't decline along a predictable path towards their death, so
- 30:14Speaker 1: it's often a series of strings that snaps. If you
- 30:17Speaker 1: think about the human being being held together like a
- 30:20Speaker 1: puppet on on many strings, it's not that the puppet
- 30:23Speaker 1: slowly crumbles at a predictable pace. It's that all of
- 30:27Speaker 1: a sudden, three strings collapse and the hand comes dangling
- 30:29Speaker 1: down and the body and medicine tries to prop that
- 30:33Speaker 1: piece up, and in doing so now two more strings
- 30:36Speaker 1: get cut, and the and the foot collapses, and when
- 30:39Speaker 1: a certain number collapses that nothing can be done. So
- 30:41Speaker 1: it's a fundamental failure of homeostasis that makes death very
- 30:45Speaker 1: hard to imagine, conceive, And of course there is an
- 30:49Speaker 1: emotional component to this, but unlike human doctors, AI doesn't
- 30:53Speaker 1: get distracted by emotion. It looks at evidence and historical
- 30:57Speaker 1: data to establish patterns. The algorithms actually do quite well
- 31:02Speaker 1: in predicting death. What is it attaching weight to? Is
- 31:05Speaker 1: it a combination of things? Is it the fact that
- 31:08Speaker 1: someone has a brain metastasis and has a slight rise
- 31:11Speaker 1: in some blood value of some solid that predicts that
- 31:15Speaker 1: this person is likely to do very badly in the
- 31:17Speaker 1: next few days. You know, as you refine it further
- 31:20Speaker 1: and further, many subtle things might start coming up that
- 31:22Speaker 1: we don't know about, and those will be the most
- 31:24Speaker 1: interesting ones. It's not just additives. That phrase. It's not
- 31:29Speaker 1: just additives. It's important to me, Kara because it connects
- 31:32Speaker 1: the dots between what sadd Arthur is saying about predicting
- 31:35Speaker 1: time of death and what Andy Schwartz was saying about
- 31:38Speaker 1: getting better decoding the human brain. They're both about understanding
- 31:42Speaker 1: systems where one plus one doesn't necessarily equal to where
- 31:46Speaker 1: unexpected results emerge from complex systems. Thinking about this makes
- 31:50Speaker 1: me physically ill. Um He's literally talking about predicting when
- 31:54Speaker 1: we will die and mind blowing. You know what, if
- 31:58Speaker 1: you could know when you will die could change how
- 32:01Speaker 1: we choose to live our daily lives. It's one of
- 32:03Speaker 1: the things I find most disturbing this whole series. So
- 32:06Speaker 1: much of how we live and how we aspire and
- 32:08Speaker 1: what we hope for is connected to our uncertainty about
- 32:11Speaker 1: when we're going to die. That and I can change
- 32:13Speaker 1: all of that. It's part of this more global line
- 32:16Speaker 1: of thinking, which is that AI kind of takes the
- 32:19Speaker 1: fun out of things well, I mean, and it also
- 32:22Speaker 1: a big part of Western culture is the Bible and
- 32:25Speaker 1: the fruit of that forbidden tree, and in Paradise Loss,
- 32:29Speaker 1: there's this warning to know, to know no more, and
- 32:31Speaker 1: there's always been this idea in history that there's something
- 32:34Speaker 1: magic about not knowing. Yeah, you know, Milton aside, Okay, okay, okay,
- 32:39Speaker 1: Nat you go. We can start to think about what
- 32:42Speaker 1: this might mean more practically. Lifespan data is directly linked
- 32:46Speaker 1: to life insurance policies. You know, this would significantly change
- 32:49Speaker 1: how much people pay. You also think about personal injury law,
- 32:52Speaker 1: which takes into account how long somebody is going to live,
- 32:55Speaker 1: so you can determine how much money they should get
- 32:58Speaker 1: for loss of quality of life. These are things that
- 33:00Speaker 1: would be greatly impacted by people knowing when they're going
- 33:04Speaker 1: to die. Right. So, Datha calls death the ultimate black box,
- 33:08Speaker 1: which I think is actually the perfect description of black box.
- 33:11Speaker 1: We know we will die, we just don't know how,
- 33:14Speaker 1: and we don't know when. And with AI it's similar.
- 33:17Speaker 1: We know we'll get a result in endpoint, but we
- 33:20Speaker 1: don't know exactly how our input factors are combined to
- 33:23Speaker 1: get there. Ironically, although AI itself is a black box,
- 33:26Speaker 1: it's helping us unpack other black boxes, death being the
- 33:29Speaker 1: ultimate one, the brain being another. And then there's the
- 33:33Speaker 1: human genome, the unique pattern of our DNA that makes
- 33:36Speaker 1: each of us us. And we've met the genome, but
- 33:39Speaker 1: there are a lot of concerns about decoding it, that
- 33:42Speaker 1: it might be a sort of Pandora's box. Well, there's
- 33:44Speaker 1: that story about the scientists in China who edited the
- 33:47Speaker 1: genes of two babies using crisper and all the ethical
- 33:51Speaker 1: concerns of creating genetic babies exactly. So I spoke with
- 33:55Speaker 1: Andy Schwartz about actual progress being made in decoding the
- 33:58Speaker 1: human genome. If you go back to the late ninety nineties,
- 34:03Speaker 1: and the race was on to discover the human genome,
- 34:06Speaker 1: and the sound bikes were as soon as we understand
- 34:09Speaker 1: all the genes, we can cure disease. So, for instance,
- 34:13Speaker 1: there was a breast cancer gene, and there was an
- 34:16Speaker 1: Alzheimer's gene, and if we just knew what those genes were,
- 34:19Speaker 1: we'd be able to eradicate these diseases. Well, it's been
- 34:23Speaker 1: twenty years now and we're just beginning perhaps to get
- 34:30Speaker 1: some sort of genome based therapies that might address some
- 34:33Speaker 1: of these And what we found is that there's no
- 34:36Speaker 1: simple cause and effect. Very rarely are there simple gene
- 34:40Speaker 1: defects correlated disease. Rather, these diseases have hundreds of genetic
- 34:46Speaker 1: basis and each of those is relatively weak, but combined together,
- 34:51Speaker 1: they generate these diseases. And so it's becomes a computational
- 34:55Speaker 1: problem and we start looking again at this as a
- 34:58Speaker 1: complex system where causality is no longer clear. And these
- 35:02Speaker 1: are the complex computational problems. We're getting better and better
- 35:05Speaker 1: at solving Sadartha himself is interested in exactly this area,
- 35:09Speaker 1: the confluence of genetics and computation. In fact, in twenty
- 35:13Speaker 1: eighteen he gave a talk at Vanderbilt University called from
- 35:17Speaker 1: Artificial Intelligence to Genomic Intelligence, and it's an area where
- 35:21Speaker 1: we're making rapid progress. The first papers are just starting
- 35:25Speaker 1: to appear there in preprint. One of them is extraordinarily interesting.
- 35:29Speaker 1: It appears to be able to predict height based on
- 35:33Speaker 1: an algorithm and genomic information to all parents tend to
- 35:36Speaker 1: produce to all children short parents and to produce short children.
- 35:39Speaker 1: But we did not have ways to predict based on
- 35:41Speaker 1: genetic information what your actual height was going to be.
- 35:44Speaker 1: The question becomes, well, how do you take the genome
- 35:47Speaker 1: and out pops height from it? If you can do that,
- 35:49Speaker 1: that means you can take a field genome in utero
- 35:52Speaker 1: and predict this person's future height. You know, based on
- 35:55Speaker 1: these first few papers that I read about this arena,
- 35:58Speaker 1: you require deep learning to do this. I wanted to
- 36:01Speaker 1: understand why this prediction of height requires deep learning algorithms,
- 36:06Speaker 1: not simple ones like Andy used to interpret Jan's brain waves.
- 36:10Speaker 1: It's not just additive you can just add up across
- 36:13Speaker 1: multiple variations in the genome and arrive at a risk score.
- 36:16Speaker 1: It's that there are interactions between genes that have to
- 36:19Speaker 1: be captured. Again, these are early days for artificial intelligence
- 36:23Speaker 1: being unleashed on genomes, but it seems to me that
- 36:26Speaker 1: complex problems of genetic architecture will soon be predictable using
- 36:30Speaker 1: these kinds of algorithms, and that ability to predict raises
- 36:35Speaker 1: huge questions for all of us. If you want to
- 36:37Speaker 1: know the height of your unborn child, but you want
- 36:39Speaker 1: to know the risk of dyslexia, those questions are almost
- 36:42Speaker 1: certainly likely to lead to extraordinarily acrimonious public conversations about
- 36:47Speaker 1: what should be done and what shouldn't be done in
- 36:49Speaker 1: terms of accessing the data, who's to store the data,
- 36:52Speaker 1: how much privacy we should have about it, and how
- 36:54Speaker 1: much it will distort human culture to have these pieces
- 36:56Speaker 1: of knowledge. Um So, if you think that clinically going
- 37:00Speaker 1: when you're going to die is is going to distort culture,
- 37:03Speaker 1: in knowing how tall your child is going to be
- 37:06Speaker 1: in the future will also distort human culture. We haven't
- 37:08Speaker 1: ever lived in a place or a space or a
- 37:11Speaker 1: time when that knowledge has been predictable from a fetus.
- 37:14Speaker 1: Artificial intelligence is giving us incredible power to see into
- 37:18Speaker 1: the future, to ask and answer questions about the generations
- 37:22Speaker 1: to come. But it is up to us, our generation,
- 37:25Speaker 1: to decide how we want to use this awesome power.
- 37:29Speaker 1: But one thing that artificial neural networks can't do is
- 37:32Speaker 1: defined principles. They can only work on classifying things that
- 37:38Speaker 1: we tell them to classify. There is still a human
- 37:41Speaker 1: telling the artificial neural network what it should be doing.
- 37:45Speaker 1: There is something very fundamental about the human brain, a
- 37:49Speaker 1: scientist's brain, a doctor's brain, and artist's brain that asks
- 37:52Speaker 1: questions in a fundamentally different manner, the why question. Why
- 37:55Speaker 1: did this happen in this person in this time? Why
- 37:59Speaker 1: does the melanoma appear in the first case? What is
- 38:01Speaker 1: the molecular basis of that appearance? The most interesting mysteries
- 38:04Speaker 1: of medicine remain mysteries that have to do with the y,
- 38:07Speaker 1: and despite being at the absolute cutting edge of medical research,
- 38:11Speaker 1: Saddatha's most important guiding principle was written in ancient Greek
- 38:15Speaker 1: over two thousand years ago. Remember the Hipocratic oath begins first,
- 38:20Speaker 1: do no harm it's maybe the single profession where the
- 38:23Speaker 1: oath of the profession is in the negative. And this
- 38:25Speaker 1: is for a reason. It's for a profound reason in medicine,
- 38:28Speaker 1: because we're intervening on bodies, because we're intervening on homeostasis,
- 38:32Speaker 1: because we're intervening on cultures. Effectively, the capacity to do
- 38:36Speaker 1: harm arises very quickly, and so the first do no
- 38:39Speaker 1: harm injunction in the Hippocratic oath is is an important
- 38:42Speaker 1: thing to keep in my own mind. Um, you know,
- 38:45Speaker 1: what are the harms that arise if I were to
- 38:48Speaker 1: start knowing my risks of future disease, not just what
- 38:51Speaker 1: advantages would I get in society. And this battle is
- 38:54Speaker 1: happening in my mind, I assume in the minds of
- 38:56Speaker 1: virtually every doctor. As we move forward into this uh
- 39:00Speaker 1: beautiful and perilous future, in a world where our choices
- 39:05Speaker 1: can create new beauty but also a new peril. It's
- 39:08Speaker 1: important that we move into that future with real care.
- 39:11Speaker 1: And it's something sad Arthur has thought a lot about personally, because,
- 39:15Speaker 1: like Sebastian Throne, he has a family history of heritable conditions.
- 39:19Speaker 1: The risk is of schizophrenic disease and bipolar disorder, and
- 39:24Speaker 1: right now the algorithms to predict this still don't exist.
- 39:29Speaker 1: As the project of sequencing lots of genomes and asking
- 39:32Speaker 1: what diseases people have matures, this data set will become
- 39:35Speaker 1: available maybe five ten years from now. I will be
- 39:39Speaker 1: past the period I suppose where that will make a difference.
- 39:42Speaker 1: But to my children and my grandchildren, it might make
- 39:44Speaker 1: a difference, and they'll have to make that decision. I
- 39:47Speaker 1: will advise them individually, and it will depend on humanistic
- 39:50Speaker 1: understanding of what an individual's desire to understand their own
- 39:54Speaker 1: risk is. There's no algorithm that predicts that understanding. As
- 40:00Speaker 1: AI advances, we're being faced with more and more urgent
- 40:03Speaker 1: ethical choices. This, in turn, may put a new emphasis
- 40:06Speaker 1: on the humanities, or, as Kai Fuey suggested, place a
- 40:10Speaker 1: new premium on personal attention, human interaction, and emotional care.
- 40:17Speaker 1: Once we give up some of the diagnostic pattern recognition
- 40:21Speaker 1: material to machines, it will be time to play. It
- 40:24Speaker 1: will be the time to play in the arena of
- 40:27Speaker 1: human therapeutics, human biology, the complexity of the human interaction,
- 40:30Speaker 1: the art of medicine. My hope is that medicine, being
- 40:34Speaker 1: more playful, will become more compassionate, more able to take
- 40:38Speaker 1: into account individuals and their individual destinies rather than bucketing
- 40:43Speaker 1: people in big categories. It means having more time to
- 40:47Speaker 1: spend with humans. You know, we are so constrained by
- 40:51Speaker 1: time that even compassion gets three minutes, we won't become
- 40:56Speaker 1: more robotic, will become less robotic as the robots and
- 41:00Speaker 1: our own What's the data describes is the holy grail
- 41:04Speaker 1: of the AI revolution. Could it allow us to be
- 41:07Speaker 1: more human, to be better doctors, more fulfilled workers, and
- 41:12Speaker 1: greater artists. Could it take routine work out of our
- 41:15Speaker 1: hands and allow us to take better care of each other.
- 41:20Speaker 1: It's a compelling vision, but as always, it has a
- 41:23Speaker 1: dark side. While most doctors are guided by their hippocratic
- 41:26Speaker 1: oath do no harm, there's no guarantee that new technologies
- 41:30Speaker 1: will stay in the right hands. The line between healing
- 41:33Speaker 1: and upgrading our bodies is thin and contested, and as
- 41:37Speaker 1: AI improves, we can begin to translate desires directly from
- 41:41Speaker 1: brain activity, modify the physical traits of our children through
- 41:45Speaker 1: gene editing, and accurately predict when we will die. In
- 41:50Speaker 1: the next episode, we ask what does all of this
- 41:52Speaker 1: mean for our future? As a species. We speak to
- 41:56Speaker 1: the world's leading thinker on these questions. You've all know
- 41:59Speaker 1: her are author of Sapiens and Homodaeus. I'm Ozveloshin. See
- 42:04Speaker 1: you next time. Sleepwalkers is a production of I Heart
- 42:19Speaker 1: Radio and Unusual Productions. For the latest AI news, live interviews,
- 42:24Speaker 1: and behind the scenes footage, find us on Instagram, at
- 42:27Speaker 1: Sleepwalker's podcast or at Sleepwalker's podcast dot com. Sleepwalkers is
- 42:34Speaker 1: hosted by me Ozveloshin and co hosted by me Kara Price,
- 42:37Speaker 1: with produced by Julian Weller, with help from Jacobo Penzo
- 42:40Speaker 1: and Taylor Chakoin mixing by Tristan McNeil and Julian Weller.
- 42:44Speaker 1: Our story editor is Matthew Riddle. Recording assistance this episode
- 42:48Speaker 1: from Joe and Luna to Brina Boden and Joseph Friedman.
- 42:51Speaker 1: Sleepwalkers is executive produced by me Ozveloshin and Mangesh Hattikiller.
- 42:55Speaker 1: For more podcasts from My Heart Radio, visit the I
- 42:57Speaker 1: Heart Radio app, Apple Podcasts, or wherever you listen to
- 43:00Speaker 1: your favorite shows.