1 00:00:00,080 --> 00:00:02,400 Speaker 1: It's time now for our Wall Street Week daily segment, 2 00:00:02,440 --> 00:00:04,720 Speaker 1: those of Wall Street Week. David Weston joins us, as 3 00:00:04,720 --> 00:00:06,800 Speaker 1: he does every day around this time. And David, there's 4 00:00:06,840 --> 00:00:09,320 Speaker 1: been so much focus this year on artificial intelligence, and 5 00:00:09,480 --> 00:00:11,680 Speaker 1: let's get beyond the hype here and really start to 6 00:00:11,680 --> 00:00:13,440 Speaker 1: talk about how this stuff. 7 00:00:13,160 --> 00:00:15,840 Speaker 2: Is actually going to be applied, specifically in the medical healthcare. 8 00:00:16,040 --> 00:00:17,759 Speaker 2: We talk with doctor Lloyd Minor, he's the data of 9 00:00:17,800 --> 00:00:19,840 Speaker 2: the stand from School of Medica about this subject. He's 10 00:00:19,840 --> 00:00:21,200 Speaker 2: got a lot of thought. You think he thinks it's 11 00:00:21,200 --> 00:00:23,960 Speaker 2: a transformative moment for medicine and healthcare. 12 00:00:25,400 --> 00:00:28,280 Speaker 3: I think the transformative aspects of generative AI can be 13 00:00:28,280 --> 00:00:32,240 Speaker 3: broken down into several categories. One is making health care 14 00:00:32,280 --> 00:00:37,960 Speaker 3: delivery more efficient and effective and equitable. Efficient in that 15 00:00:38,640 --> 00:00:42,479 Speaker 3: physicians other healthcare providers will have access to information in 16 00:00:42,560 --> 00:00:46,840 Speaker 3: real time and in ways that actually inform decisions about 17 00:00:46,880 --> 00:00:50,920 Speaker 3: the care of individual patients. Effective and that hopefully we'll 18 00:00:50,960 --> 00:00:55,520 Speaker 3: eliminate a lot of unnecessary testing, a lot of procedures 19 00:00:55,640 --> 00:00:59,440 Speaker 3: or activities that don't directly link to the well being 20 00:00:59,480 --> 00:01:02,840 Speaker 3: of the patient. And equitable because it will provide access, 21 00:01:02,920 --> 00:01:06,920 Speaker 3: particularly access to specialty services that today are not widely 22 00:01:06,920 --> 00:01:09,199 Speaker 3: distributed in our country and certainly in the world. 23 00:01:09,240 --> 00:01:11,399 Speaker 2: I wonder about the distributional aspect of that of the 24 00:01:11,480 --> 00:01:13,520 Speaker 2: delivery that you refer to, because there is an issue 25 00:01:13,560 --> 00:01:16,679 Speaker 2: in this country that not all people get equal access 26 00:01:16,880 --> 00:01:19,920 Speaker 2: to healthcare. We certainly saw it during the pandemic. Is 27 00:01:19,959 --> 00:01:23,280 Speaker 2: this an area where I could make a particularly disproportionate 28 00:01:23,280 --> 00:01:25,080 Speaker 2: contribution to the healthcare system? 29 00:01:25,400 --> 00:01:27,920 Speaker 3: Yes, David, I think it really is an area where 30 00:01:28,120 --> 00:01:33,120 Speaker 3: it can have a disproportional, disproportionate favorable effect, so we know. 31 00:01:33,200 --> 00:01:37,840 Speaker 3: For example, an example I like to draw is in dermatology. 32 00:01:37,959 --> 00:01:41,280 Speaker 3: There are dermatologists, certainly in big metropolitan areas like here 33 00:01:41,280 --> 00:01:43,480 Speaker 3: in New York, but there are certain regions of the 34 00:01:44,560 --> 00:01:48,320 Speaker 3: country that don't have dermatologists or that have them, and 35 00:01:48,360 --> 00:01:50,280 Speaker 3: they're so busy that it's hard to get an appointment. 36 00:01:50,960 --> 00:01:56,440 Speaker 3: Our computer scientists, working with our dermatologists, developed an algorithm 37 00:01:57,040 --> 00:01:59,960 Speaker 3: that will from a picture taken of a skin leasion 38 00:02:00,440 --> 00:02:03,800 Speaker 3: can as accurately make a diagnosis of whether that's cancer 39 00:02:03,920 --> 00:02:07,400 Speaker 3: or not as a trained dermatologist looking at the same lesion. Now, 40 00:02:07,560 --> 00:02:10,560 Speaker 3: the dermatologist does better when they're actually examining the patient 41 00:02:10,639 --> 00:02:13,120 Speaker 3: than just a picture. The point is that offers the 42 00:02:13,160 --> 00:02:16,520 Speaker 3: opportunity for a physician in rural America that may not 43 00:02:16,639 --> 00:02:19,079 Speaker 3: have access to a dirt dermatologist for a couple of 44 00:02:19,160 --> 00:02:23,600 Speaker 3: hours away, to be able to accurately screen a lesion 45 00:02:23,639 --> 00:02:25,799 Speaker 3: and know whether or not that patient needs to see 46 00:02:25,800 --> 00:02:26,840 Speaker 3: a dermatologist or. 47 00:02:26,800 --> 00:02:29,040 Speaker 2: Not, which may lead to a critical distinction. You have 48 00:02:29,120 --> 00:02:31,320 Speaker 2: a doctor involved in all of that. It's not that 49 00:02:31,480 --> 00:02:34,760 Speaker 2: general AI replaces the doctor. It's not substitution doctor, but 50 00:02:34,880 --> 00:02:38,040 Speaker 2: augmentation allows the doctor to do a better job, faster 51 00:02:38,320 --> 00:02:39,520 Speaker 2: and more broadly distributed. 52 00:02:40,000 --> 00:02:43,640 Speaker 3: That's right. And when we started deploying AI in the 53 00:02:43,639 --> 00:02:47,920 Speaker 3: interpretation of radiology images, which was several years ago, five six, 54 00:02:48,000 --> 00:02:52,320 Speaker 3: seven years ago, there were rumors going out that there 55 00:02:52,360 --> 00:02:56,200 Speaker 3: weren't going to be radiologists anymore, and that was of 56 00:02:56,200 --> 00:02:59,080 Speaker 3: course false. And I think there will always be a 57 00:02:59,160 --> 00:03:02,760 Speaker 3: need for radiology, and there will be radiologists who know 58 00:03:02,800 --> 00:03:06,239 Speaker 3: how to use AI and do deploy it in their practices, 59 00:03:06,720 --> 00:03:08,919 Speaker 3: and there will be those that don't. And in the end, 60 00:03:09,000 --> 00:03:12,240 Speaker 3: those that do and use it responsibly are going to 61 00:03:12,240 --> 00:03:15,760 Speaker 3: provide the better care and be able to deliver care 62 00:03:15,800 --> 00:03:17,639 Speaker 3: more effectively. We've been talking about delivery. 63 00:03:17,880 --> 00:03:20,200 Speaker 2: Let's talk about another aspect of health care, and that 64 00:03:20,320 --> 00:03:24,080 Speaker 2: is actually underlying services, and particularly something called synthetic biology, 65 00:03:24,120 --> 00:03:26,560 Speaker 2: which you have talked and written about. Tell us about 66 00:03:26,600 --> 00:03:28,840 Speaker 2: synthetic biology and what that could do to healthcare. 67 00:03:29,480 --> 00:03:33,960 Speaker 3: Synthetic biology refers to engineering life. It involves taking a 68 00:03:34,040 --> 00:03:40,080 Speaker 3: cell altering it to do another function. And already today 69 00:03:40,280 --> 00:03:45,240 Speaker 3: we're using synthetic biology, for example, in therapies for cancer, 70 00:03:45,280 --> 00:03:51,120 Speaker 3: where we engineer a person's own immune system cells in 71 00:03:51,200 --> 00:03:53,760 Speaker 3: order to attack cancer in ways that those cells wouldn't 72 00:03:53,760 --> 00:03:56,839 Speaker 3: otherwise be able to do. But in the future we'll 73 00:03:56,840 --> 00:03:59,920 Speaker 3: be able to engineer life to do other things as well. 74 00:04:00,440 --> 00:04:05,200 Speaker 3: It can have enormous implications and sustainability, for example, where 75 00:04:05,240 --> 00:04:08,320 Speaker 3: we may be able to engineer life to do better 76 00:04:08,600 --> 00:04:13,840 Speaker 3: carbon capture or make micro organisms to solve plastics. But 77 00:04:14,480 --> 00:04:19,279 Speaker 3: with this opportunity to change cells to alter their metabolism, 78 00:04:19,440 --> 00:04:25,080 Speaker 3: their gene expression, goes the opportunity to really rethink problems 79 00:04:25,080 --> 00:04:27,160 Speaker 3: that we haven't been able to confront in the past. 80 00:04:27,560 --> 00:04:31,560 Speaker 2: What you're describing is a very different world of medicine 81 00:04:31,600 --> 00:04:33,839 Speaker 2: and of research for that matter. You are dean at 82 00:04:33,839 --> 00:04:37,080 Speaker 2: the Stanford Medical School. How does this change how you 83 00:04:37,200 --> 00:04:40,320 Speaker 2: train the next generation of physicians and for that matter, 84 00:04:40,360 --> 00:04:41,400 Speaker 2: of research scientists. 85 00:04:41,680 --> 00:04:44,320 Speaker 3: I think it's going to have a radical effect on 86 00:04:44,440 --> 00:04:49,240 Speaker 3: the way we educate physicians, physicians, scientists, and scientists. When 87 00:04:49,320 --> 00:04:52,880 Speaker 3: we an institutional like Stanford, when we have the privilege 88 00:04:52,880 --> 00:04:55,520 Speaker 3: of working with students for four, five, six, how many 89 00:04:55,560 --> 00:05:00,560 Speaker 3: other years, we're actually trying in our training programs to 90 00:05:00,560 --> 00:05:05,920 Speaker 3: prepare them for thirty forty years of productive life after that. 91 00:05:06,920 --> 00:05:11,120 Speaker 3: So we're training learners more than we are imparting facts 92 00:05:11,120 --> 00:05:14,680 Speaker 3: of the moment or knowledge of the moment. Generative AI 93 00:05:15,120 --> 00:05:19,159 Speaker 3: large language models, particularly as they incorporate other representational data 94 00:05:19,279 --> 00:05:24,719 Speaker 3: like images, that radically changes the type of knowledge and 95 00:05:24,800 --> 00:05:28,520 Speaker 3: ability that people being trained today need to have. There's 96 00:05:28,600 --> 00:05:32,760 Speaker 3: less emphasis, there will be less emphasis on memorization, more emphasis, 97 00:05:32,839 --> 00:05:36,800 Speaker 3: I think, on truly understanding and in particular understanding what 98 00:05:36,839 --> 00:05:40,000 Speaker 3: these models are doing and how they can be responsibly 99 00:05:40,040 --> 00:05:41,600 Speaker 3: deployed and trained. 100 00:05:42,080 --> 00:05:44,360 Speaker 2: If you are going to guess, do you believe generative 101 00:05:44,680 --> 00:05:48,320 Speaker 2: in the end will make medical practice more attractive or 102 00:05:48,400 --> 00:05:51,720 Speaker 2: less attractive to young people coming into field, Well, I. 103 00:05:51,640 --> 00:05:54,440 Speaker 3: Hope it makes it more attractive because the more that 104 00:05:54,480 --> 00:06:00,320 Speaker 3: we can responsibly deploy technology and the delivery of healthcare more. 105 00:06:00,360 --> 00:06:03,240 Speaker 3: We should be able to get physicians and other healthcare 106 00:06:03,279 --> 00:06:06,160 Speaker 3: providers back to what they really want to do, and 107 00:06:06,200 --> 00:06:09,719 Speaker 3: that is interacting with patients during critically important moments in 108 00:06:09,760 --> 00:06:14,680 Speaker 3: their lives. Unfortunately, in the past, technology when it's initially implemented, 109 00:06:14,880 --> 00:06:19,640 Speaker 3: has oftentimes been a barrier to healthcare providers interacting with patients. 110 00:06:19,960 --> 00:06:23,279 Speaker 3: I think those barriers are about to be dramatically lowered 111 00:06:23,560 --> 00:06:25,480 Speaker 3: because of the improvements in the technology. 112 00:06:25,600 --> 00:06:27,600 Speaker 2: The world you describe as an attractive world a lot 113 00:06:27,600 --> 00:06:30,080 Speaker 2: of benefits from general AI. Where are there risks? There 114 00:06:30,080 --> 00:06:33,520 Speaker 2: are always risks. With any powerful tool, there. 115 00:06:33,360 --> 00:06:36,880 Speaker 3: Are many risks, and a couple of weeks ago, the 116 00:06:36,880 --> 00:06:41,560 Speaker 3: Biden administration published an executive order on sort of the 117 00:06:41,560 --> 00:06:46,560 Speaker 3: future of AI, and a key component in that statement 118 00:06:47,120 --> 00:06:51,080 Speaker 3: is a focus on patient privacy. We absolutely have to 119 00:06:51,200 --> 00:06:53,880 Speaker 3: keep that in mind, and that also means data security, 120 00:06:54,520 --> 00:06:57,880 Speaker 3: which is linked to AI and other developments and technology. 121 00:06:58,480 --> 00:07:01,040 Speaker 3: We also have to keep in mind that these models 122 00:07:01,080 --> 00:07:03,520 Speaker 3: are only as good as how they're trained, the data 123 00:07:03,600 --> 00:07:06,320 Speaker 3: used to train them, and if that data is biased, 124 00:07:06,839 --> 00:07:09,840 Speaker 3: then the outcome the predict predictions of the model is 125 00:07:09,880 --> 00:07:12,280 Speaker 3: going to be biased as well. So those are some 126 00:07:12,320 --> 00:07:13,840 Speaker 3: of the risks that we have to keep in mind. 127 00:07:14,120 --> 00:07:16,840 Speaker 3: I think we can responsibly manage. I think we have 128 00:07:16,960 --> 00:07:20,840 Speaker 3: to responsibly manage those risks, but first comes the awareness 129 00:07:20,840 --> 00:07:21,760 Speaker 3: of what the risks are. 130 00:07:23,120 --> 00:07:25,640 Speaker 2: That was like a Lloyd minor dean in the Stanford School. 131 00:07:25,640 --> 00:07:27,120 Speaker 2: But I don't know what you Ryan. I find it 132 00:07:27,120 --> 00:07:28,880 Speaker 2: pre encouraging. So I think there's good stuff that I 133 00:07:28,880 --> 00:07:29,400 Speaker 2: could come out of that. 134 00:07:29,480 --> 00:07:31,400 Speaker 1: Well, that's what I'm waiting for here, because I mean, 135 00:07:31,480 --> 00:07:33,440 Speaker 1: so much of the hype around AI's just that hype. 136 00:07:33,520 --> 00:07:35,720 Speaker 1: Or it's about how you can sell me ads or something. 137 00:07:36,200 --> 00:07:38,040 Speaker 1: Give me something practical and useful to my life. 138 00:07:38,080 --> 00:07:39,640 Speaker 2: Yeah, make you healthier, that'd be a good thing.