How AI is Transforming Healthcare and Restoring Humanity | Dr. Robert Wachter - EP 742

Coming up next on Passionstruck. I'm more optimistic about healthcare AI than I am about the AI in the rest of our lives, because in healthcare, I can't find a doctor, a nurse, or a patient who says the system is just perfect. And we don't need to screw around with it. It's delivering the high quality, safe, accessible care at an affordable price. Nobody thinks that. And I don't think we can deliver on what patients need unless we have the assistance of a technology like AI. So it really is both the quality of the technology and the desperate need that we have to make things better. Welcome to Passionstruck. I'm your host, John Miles. This is the show where we explore the art of human flourishing and what it truly means to live like it matters. Each week, I sit down with change makers, craters, scientists, and everyday heroes to decode the human experience and uncover the tools that help us lead with meaning, heal what hurts, and pursue the fullest expression of who we're capable of becoming. Whether you're designing your future, developing as a leader, or seeking deeper alignment in your life, this show is your invitation to grow with purpose and act with intention. Because the secret to a life of deep purpose, connection, and impact is choosing to live like you matter. Hey, friends! Welcome back to episode 742 of Passionstruck. Over the past several episodes in our Life Behind the Scripts series, we've been exploring how transformation happens across different layers of the human experience. With Joan London, we looked at identity. What happens when the roles that once defined you no longer fit? With Harvard professor Leslie John, we explored authenticity. How revealing more of yourself reshapes connection and belonging. With biohacking pioneer Dave Asprey, we turned to biology. How your brain, body, and nervous system influence the person you're capable of becoming. And in my conversation last Thursday with Dr. Tara Nourula, we focused on resilience. How agency, connection, and mindset shape not just how long we live, but how well we live. Today, we zoom out even further, because there's another force reshaping our lives. Not just individually, but systematically. Technology. Artificial intelligence is already transforming how we work, communicate, and make decisions. And nowhere may its impact be more profound than in healthcare. The domain where technology intersects directly with human vulnerability, trust, and survival. My guest today is Dr. Robert Wachter, physician, professor, chair of the Department of Medicine at University of California, San Francisco, and one of the world's leading voices on healthcare innovation. We're discussing his new book, A Giant Leap. How AI is transforming healthcare and what that means for our future. Dr. Wachter has spent decades studying how medical systems evolve from electronic health records to patient safety to digital medicine. And he offers a uniquely balanced perspective on both the promise and the risks of AI. In this conversation, we explore how AI could dramatically improve access to care. Why burnout and administrative overload have reached a breaking point. The surprising ways AI may restore humanity to medicine. The risks of bias, misinformation, and deep fakes in high stakes decisions. And how patients themselves may become far more empowered participants in their own health. This is not just a conversation about technology, it's a conversation about trust, responsibility, and the future of human care. Before we dive in, a quick ask. If this episode resonates with you, share it with someone who cares about the future of healthcare. You can also watch the full conversation on our YouTube channels. And if you haven't yet, leaving, a rating, or review on Apple Podcasts or Spotify helps more people find these conversations. Now let's dive into my conversation with Dr. Robert Wachter. Thank you for choosing Passion Struck and choosing me to be your host and guide on your journey to creating an intentional life that matters. Now, let that journey begin. There's something about March that makes you want to reset your space. For me, that started in the kitchen, clearing out what I didn't need and upgrading what I use every single day. That's when I brought in Careaway. What I love is how simple it makes everything. The ceramic coating means food just lifts right off, so I'm using less oil. Clean up takes seconds and cooking actually feels enjoyable again. But it's not just performance, it's intentional design. The storage system keeps everything organized and my cabinets finally feel calm instead of chaotic. And I've become a lot more mindful about what I bring into my home. Careaway is third-party tested and made with high standards, which just gives me more confidence in what I'm using every day. It's one of those upgrades you feel immediately. Careaway Cookware Set is a favorite for a reason. It can save you up to $230 versus buying the items individually. Plus, if you visit careawayhome.com slash Passion Struck, you can take an additional 10% off your next purchase. This deal is exclusive for our listeners, so visit careawayhome.com slash Passion Struck or use code Passion Struck at checkout. Careaway, non-toxic kitchenware, made modern. I am so excited today to welcome Dr. Robert Wachter to Passion Struck. Welcome, Bob. How are you today? I'm great, John. Thank you so much for having me. Well, I've read some interviews you've done and you've joked that you were born in Brooklyn before it was trendy. How did growing up in that second generation Jewish family shape how you see work, success, and responsibility? I was born there and then we moved out to Long Island. Neither of my parents went to college. And I guess the funniest story was my dad was really an admirer of doctors. He ran a small business and I remember going, he would go to cocktail parties on Friday night and he'd be dressed normally and then I'd look down and he'd have his garage door open on his belt. I said, Dad, what's that? He said, there are a lot of doctors at the party. They have beepers. I have nothing. So I think it's shaped me as having parents who didn't go to college. They very much wanted me to do well, to make a contribution. I think they saw medicine as the highest calling. For me, it was a little bit interesting because I was also a politics junkie and had no idea how I'd be able to shape a career that had my interest in sort of the way systems worked and people interact with each other and also be a doctor. And I feel incredibly fortunate things have worked out pretty well. Yeah, I understand that you never thought you'd be able to combine political science with medicine, but it's turned out to happen over time. How did that end up occurring? I met a mentor in medical school at the University of Pennsylvania, a guy named John Eisenberg, who is an MD and an MBA from Wharton. And John was incredibly talented as a physician, as a teacher, as a researcher. He was charismatic. He was larger than life. And what he studied, I thought research in medicine was genes and Bunsen burners and test tubes. And that was, I was not that interested in or nor that good in basic science. But what John studied was the system, how we organize ourselves, how the payment system influences care, why are so many people uninsured, and how do we fix that? So it was really through John that I said, wow, there is a career here as an academic physician, where you study the system of healthcare. And that was completely in sync with my interest in policy and politics and economics. And I realized, wow, you can be a doctor part time and a researcher part time. And what you can study is this system and how to make it better. And that's given me plenty to do over the last 40 years, because the system really needs a lot of reform. Yes. So what we're going to be discussing today is your new book, a giant leap, how AI is transforming healthcare, and what that means for our future, which I think is a really important conversation, because I can't think of AI transforming any industry as much as it could potentially transform healthcare. And Lord, do we need it? But you open the book with, yeah, you open the book with the Mayo Clinic digital twin, Dr. Morris watching his own AI double explain a diagnosis to a patient. When you first saw that, what did it stir in you emotionally? I found it thrilling. And because I recognize that the system that we currently have has such an extraordinarily hard time providing the access that patients need. And I don't just mean patients who are in rural areas or patients who are in countries that don't have enough primary care doctors or cardiologists or oncologists, even patients who are in a city like San Francisco, it's just very hard to get an appointment with a primary care doctor or you have a medical issue and you don't really think you need to see a doctor, but you need to get good information. And so seeing the CEO of the Mayo Clinic showed me this video of a doctor talking to a patient and the doctor was empathic and quite human answering questions. And then the big reveal, of course, was those the real doctors walked in behind this digital twin. And so I found it exciting and really opens up the idea that AI could scale the knowledge of individual doctors or of places like UCSF or where I work or the Mayo Clinic. After I was done being thrilled by that, I got scared because in some ways the dark side of that is the idea of deep fakes. And that's probably the thing I worry about the most in healthcare, maybe I worry about the most generally in AI, which is the capacity to create a digital twin of me, which I think I would trust, talking to you about your health problems also means we have the capacity to create a digital twin of me telling you you should drink lattes to treat your cancer or you can think of all the kind of absurdities and it looks like me speaking. And so like a lot of technology, it's a double edge sword, but I think the net positives are pretty exciting. I was listening earlier this morning to a keynote that you gave on this topic and during that keynote and in the book you referenced Ernest Hemingway's gradually then suddenly. And I know for a lot of people, electronic health records were supposed to be a healthcare's sudden moment, but they weren't. Why does AI feel like it might just be? I think there are a couple of reasons. Hemingway is lying about how does the mango bankrupt gradually and suddenly feels perfect because we certainly have the gradually part down pat. If you think about it, healthcare may be with the exception of education, healthcare is the last industry to be transformed by technology. Just think about the way you access entertainment on Netflix or Amazon Prime or whatever. In San Francisco, I take a Waymo, I take a car with no driver about once a week. I'll sit in the back seat, sing Springsteen at the top of my lungs and sometimes take a nap and completely trust this thing will get me to a place that's safe. Whereas in healthcare, we still are the biggest users of fax machines. We finally ditch the beepers, but we're the biggest users of fax machines in the universe. It has been gradually. Why suddenly? I think two things. One is this AI is really quite remarkable and it has a talent that no prior AI had, which is that it can read unstructured data or hear unstructured speech and understand it or quote understand it's not human, but take it in, process it, and then provide outputs that are really remarkably human like. In medicine, that's quite important. We had AI up until before three years ago. It could analyze computable data. It could take your hemoglobin number or your creatinine number or all that and make sense of that, but it couldn't analyze my note about your visit. Now this new AI can. The capacity of it to do really amazing things is there. But I think the probably the more important reason is that the healthcare system desperately needs it. And that's why I'm in some ways I'm more optimistic about healthcare AI than I am about the AI in the rest of our lives because in healthcare, I can't find maybe you know some folks, John, but I can't find a doctor, a nurse or a patient who says the system's just perfect. We don't need to we don't need to screw around with it. It's delivering the high quality, safe, accessible care at a an affordable price. Nobody thinks that and I don't think we can deliver on what patients need unless we have the assistance of a technology like AI. So it really is both the quality of the technology and the desperate need that we have to make things better. If you look at healthcare over the last 20 years, the way we have dealt with unmet needs is we've just hired more humans. See, not enough doctors. All right, we need will hire nurse practitioners or physicians assistants. Patients with cancer have incredibly challenging pathways where their care needs to be coordinated. Okay, we'll hire a boatload of care coordinators. Dealing with insurance companies is complicated. We'll hire a thousand people in my hospital to deal with the insurance companies and write authorization requests and all that. It's we can't do that anymore. We can't afford it and we can't even find the humans anymore. So it's really the combination of really a breathtaking technology, but also an industry that absolutely needs the help. I don't think I would have thought a decade ago I would ever say this, but I think the healthcare system could actually look at what I've experienced at the VA as an upgrade. And where I'm going with this is, yeah, a funny thing to say, but we have this system called my healthy vet and in it, I as the patient can look at my complete medical record. I can look at all my lab results. I can see images that are taken, even which is great when you have a wife who's a PCP because you can see the tests that were done from MRIs to blood tests, etc. But what I really like about it is when I go to community care outside of the system, you can also upload all of those images and files. And so your primary care provider within the VA has access to all of that. So they can clearly see the complete patient part. But that is such an anomaly compared to what most people experience where they can't see their record go from one department to another and they're certainly not seeing their test results. I guess let me just say, John, that I think that has improved a fair amount. Most people in the United States now get care outside of the VA get care in systems that have an electronic health record made by a Wisconsin company named Epic. And if you have Epic, you have my chart, which is a version of my healthy vet. It is a patient portal. And the problem is the VA is one uniform system across the country. And so if you go to an ER a thousand miles away, if it's a VA ER, the information will be there. In Epic, it probably won't be or it's a little harder to access. But in both cases now, I think the patient portal is a real step forward and has democratized care. The problem as we think about AI is we democratize care but didn't give people any of the tools that they need really to use it. So patients now are seeing, well, my magnesium is low and my TSH is high and my EKG is abnormal. With absolutely no context or no information to help them interpret those things. And AI will help them do that. As your wife will tell you, I'm sure, the other thing that the patient portal has done is in addition to having people have all this access to this information but really no tools to deal with it, or even basic tools to schedule an appointment the way I can schedule a haircut or a dinner reservation, it's given them just enough to confuse them or give them the illusion of democratization but none of the tools. And Epic and the other EHR vendors have done something that I think everybody thought was helpful, which is put a little button on the bottom of the screen that says send a message to your doctor. And so patients being rational human beings do that. They click that button. And so someone like your wife not only has her eight or nine or 10 hours a day of seeing patients, she comes home at night and after dinner or putting the kids to sleep, now has two hours on average of work, not only completing her charts, but answering on average 50 of these questions a night. It's not doable. And I think AI has the promise of both interpreting all of this data for patients and I think answering a fair number of the questions. So it's really on top of this democratization, we just need tools like this to make the whole thing work. Yeah. So Bob, I understand you were once optimistic that electronic health records would transform medicine. And then what they've ended up doing as you're describing is they become one of the biggest drivers of physician burnout because it's like Epic brought every single attorney, HR specialist and quality control person into the room and asked them what questions should we ask a doctor to answer. What checkboxes should we put on the screen to make sure the doctor is spending all of his or her time looking down at the computer checking rather than looking you in the eye. Yeah. And it leads to this question. You write that healthcare's digital failures had shockingly little to do with the technology itself, but more about what I'm describing. What were the real failures and how do we prevent them from happening with AI? It's a terrific question. The thing that the epiphany I had when I wrote a book 10 years ago called The Digital Doctor about the electronic health record and this phase was that this was not so unusual. The history of what are called general purpose technologies. These are new technologies that transform everything about the nature of work. And so examples of this were electricity or the steam engine or the automobile or the computer or the internet and now AI. The tradition is something called or the paradigm is what's called the productivity paradox of IT of information technology. And the paradox is it looks great in the demos. It looks great on the PowerPoint slides and you bring it in and it does not achieve what you had hoped for and been promised. And the reason is that two reasons. One is the technology is not very good in version one and has to get better as the companies iterate on it. But the most important thing is that industries and human managers typically bring the technology in and don't do the hard work of transforming the way work is done. And that just means we're human. We don't have the imagination that we need to understand what our work with technology is going to look like until the technology is actually in our world. This is an old problem in technology. Henry Ford was reputed to have said if I asked people what they wanted they would have said faster horses. They had no ability to think about what cars would do for them or their lives until there were cars. And I think the same thing is true here. The electronic health record we brought it in. We turned it on. We didn't really think for a moment about OK what's going to happen when patients have a portal with all this data but they don't understand it. And we have a little button that says send a message to your doctor. Of course they're going to click that button. We had no plans for what does that do for the workforce the workflow. How does payment work all of those sort of things. Some ways it was a natural set of mistakes and our optimism was in some ways misplaced. I think from AI a few things have happened. One is we've learned the importance of anticipating to the degree that humans can how to change the nature of the work. We've built up governance structures and health care organizations and in Dr. Zawis to figure out how to vet the tools we're not as gullible as we were to the pitches. We want to see evidence about whether it works. We want to be sure that clinicians were involved in development of these tools. I think most of our organizations have built the kind of governance that we needed for other digital tools but now is more ready to deal with AI. The AI is just easier to deal with. You don't need to know a whole lot of technology in order to use it effectively. If you think about the way you use GP to your Gemini you don't need to be a computer nerd to use it effectively. So I think we're and as I say I think the needs of the health care system are so vast that I think we're likely to embrace it. I guess the final thing is there's some politics here and the politics are that if the incumbents particularly the most powerful ones fear for their jobs and of course everybody's a little bit worried about what's going to happen to my job with AI but if the incumbents fear for their jobs and they won't welcome AI the way it would be great for them to. But in health care I think for the foreseeable future the jobs of doctors and nurses are going to be safe. There's a lot of administrative people and back office people and they may I think be challenged in terms of their jobs but I think we're being smarter about it this time. I think most doctors and nurses are going to welcome it into their world because they know how hard their job is and how they can't deliver to the patients what they want to deliver without it. I'd say maybe one final thing is in a technology this complex and in some ways this fraught you want to start with low-hanging fruit. You don't want to start on the hardest problems and in medicine that means don't start with the AI making a diagnosis for you or recommending a complex treatment for you. Start with a digital scribe where the AI will record my conversation with you as a patient and create my documentation so I can look you in the eye when you're coming in my office. Start with summarizing the patient's old chart. One out of five patients has a medical record longer than Moby Dick, longer than 600 pages. I can't possibly review that in detail in three minutes before I see you. So that is something AI can do well. Start with the things that are quick wins to get the kind of buy-in that you need. I think those are all lessons that we didn't understand with the electronic health record and I think we're doing a better job this time with AI. I want to pause for a moment. One of the core ideas in this series is that rewriting your life isn't only about personal insight, it's also about navigating the systems you live within. Your health system, your work environment, your technology landscape, your community. These forces shape what's possible long before we consciously notice them. On theignitedlife.net I'm sharing companion reflections and articles for each episode in this series, designed to help you think more deeply about your own life. Because awareness creates insight, but intentional action creates agency. If you want to explore the reflections for this episode, you can visit my substack at theignitedlife.net. Thank you for supporting those who support the show. Now a quick break for our sponsors. You were listening to Passionstruck, airing on the Passionstruck network. Now let's return to the discussion with Dr. Robert Wachter. I wanted to just make a couple comments. I spent most of my career before I got into what I'm doing now in technology. I was the CIO of a Fortune 50 company. I understand what you're saying about how most people equate technology change is going to radically revolutionize whatever it is they're doing. What everyone fails to realize is it's the cultural change, the people change, that is the biggest burden that any of these implementations have to overcome. After I left that Fortune 50 company, I went to a company that was in the healthcare space and I had been a long term employee at Lowe's. We had built this system called Total Closed Loop where we were trying to give a Lowe's customer a total closed loop experience regardless of whether they came in through the store, through the website, through the call center. We were trying to implement this in healthcare and as we got into it, we realized how problematic it was because the cornerstone of any of these transformations is trying to get your arms around data or what we call metadata in IT and that is so difficult to do in the healthcare ecosystem because the sources of data are so distributed that there's not a single one. Hopefully that's changing but that was one of the most difficult things to get your arms around. So getting a little bit better but not still hard and also if you think about it in a corporate world, your interactions are mostly with an individual consumer and if you convince them that you have a product that's terrific and the right price, they'll buy from you the number of middlemen in healthcare that you have to deal with in this as you try to close the loop. It also needs to include the insurance company, your employer, the government. It's just the level of complexity is so much higher than people are used to if they've been in more consumer-facing industries. So since he just left his illustrious career, I've got to bring Warren Buffett into this conversation because you quote him saying that healthcare is the tapeworm of the US economy. So it begs the question, does AI meaningfully change that equation or does it just digitize the parasite? That's a great question. I love that quote because it's absolutely true if you think about it. I'm old enough to remember when healthcare was 12% of the GDP and people said this is unsustainable. It's sapping up so much of our resources. Well, they're obviously wrong because it's about 20% today and at 20% of GDP, it just takes a larger and larger portion of business profits or businesses ability to invest in to do R&D in government funding and therefore government's ability to either keep taxes low or house the homeless or provide decent public education. All of those things are true. Will AI save us money? I think in some ways that one of the unanswered questions, I'd say if you looked at the last 20 or 30 years of healthcare, you would be skeptical that it will. In every other industry, when we innovate with technologies, we usually lower costs and lower prices. In healthcare, when we innovate with technology, we somehow figure out a way of not lowering prices and or lowering overall costs. Now, part of that is the demand curve is elastic. And what I mean by that is as Lasik's surgery for your eyes or replacing your hip becomes more affordable than more people want it and more people get it. And therefore, even if the net cost of that procedure goes down, the overall cost of society goes up. I think there's another difference, which is health insurance. Nobody really cares. Nobody talks about the percent of our GDP that we spend on computers or on Ubers or on restaurants. People make purchasing decisions, so they make their own choices about those things. Whereas in healthcare, because in so many cases, you're not paying out of pocket, it's coming out of some public pool, whether your businesses or government. The ways it could decrease costs. So I'm a little hopeful. The ways it could decrease costs are two major costs are labor. And so if we need fewer humans around to do the things that we need to do, or if you're able to get care yourself through an app rather than needing to see a doctor, that could lower costs. The second is that it's been said that the most expensive piece of technology in a hospital is the doctor's pen. Now, it's not the doctor's pen anymore. It's the doctor's fingertips, because what doctors make is about 8% of the healthcare dollar, but the decisions that doctors make is 80% of the healthcare dollar. So a way it could lower costs is if the AI gives me decision support and says, oh, it looks like you're about to order drug A for this patient's cancer or cholesterol or diabetes, it turns out drug B works just as well and is half the price. Don't you want to do drug B, doctor? Or maybe you have to do drug B, doctor. That could lower costs. On the other hand, I could also see a scenario where it says drug B is a little bit better than drug A and costs an extra $50,000 a year, and we recommend you use the more expensive drug. So a lot of that is going to be the devil's, it's going to be in the details of how do those recommendations get framed, whose values are underlying the recommendations, because sometimes that's a tough decision about whether you recommend the more expensive therapy for a small marginal gain, and the AI is going to have embedded in it some human's logic or some business's logic about what recommendation it gives you. Thank you for explaining that, and I want to get more into the power in the pitfalls, but because this conversation has been a little bit deep, I thought I'd bring in a little bit of humor. So I understand that you once played Elton John in front of 2,000 people and we're a college mascot. If I have my facts correct. How important has humor been in surviving in a high-stakes professional life like you have? I think it's absolutely critical. I completely believe that it's vital not to take ourselves too seriously, to try to find joy in work, even the work I do. I just finished 10 days on medical service at UCSF, taking care of very sick patients. Several of my patients died incredibly tragically, and yet you need to figure out, I don't think you can survive the work or be a very effective if you're not a whole human being. And a whole human being, I think, includes being able to laugh at the things in life that are funny or ironic, and there are many of them. For me, also, I'm chair of a very large Department of Medicine at a spectacular place. I have a thousand doctors that report to me. I never saw myself as a leader because I always thought I was too goofy, and you've given two examples of my goofiness. And now I've realized, and I had a little imposter syndrome the day my first day as chair, and then I realized actually the job of leadership today, a lot of it is about authenticity and accessibility and empathy and understanding where people are at. And I think humor is absolutely critical in being the kind of leader that people find accessible. And I had a view of leadership of these people chiseled into Mount Rushmore, and I just don't think that's effective leadership anymore. So I'm going to just detour one more time from our topic. A lot of what I discuss is around the topic of human flourishing, and I've just been in over a year writing a book on mattering and how to restore mattering into society because I think it's been slowly being stripped from us. And one of the biggest areas that mattering seems to be stripped is in the workforce. I mean, you can see this in employee engagement surveys. So being in your position in charge of so many doctors, what do you see as one of the clear gaps in the workforce that might be causing people to feel this loss of significance? Yeah, and it's shocking to me when I feel that in my world, because I can't imagine a job or a workplace where things are more set up for flourishing and for people finding meaning in the work. And academic medical centers are remarkable animals because the law schools don't run a law firm, the business schools don't run businesses. We take care of patients, we teach, we do research. It's fundamentally interesting and important where it's being sapped from our workforce. And this is an area where I'm hopeful that AI will help, is people spend so much of their time doing kind of bureaucratic paperwork that adds no value to the patients. And we all realize that this is not, it's not useful time. And over time, it just saps people's energy. It takes them away from the meaning that brought them into the field in the first place. I'm hopeful here that AI will make a big difference. I think that one of the things it does really well is some of the bureaucratic hassles. It's been estimated that about 30% of healthcare dollars are spent on kind of administrative paper pushing that really doesn't add, and I'm not arguing that there should be no regulations there. There needs to be. What we do is very high stakes. We got to get it right. But it is just layered and layered. And the amount of paperwork that we do, it can overwhelm you in your day. So that's the part I'm hopeful about, that this will take some of that away. If you can, if for your wife as a primary care doctor, if instead of spending two or three hours a day answering emails or writing to insurance companies or keeping her head down, checking boxes, she can take that time and repurpose it by looking at you in the eye and really being present for her conversations. I'm guessing her work life gets substantially more satisfying. That would be fantastic. Yeah, for all of us. For her and so many other doctors. Yeah. So I want to talk about some of the potential pitfalls of AI. So I wanted to bring in a scenario. I used to live in Austin, Texas, before I moved to Florida, where I live now. And a good friend of mine is a radiologist. And I remember he would, he was a partner in a pretty big practice. But like once a month or every six weeks, he would be on call over the weekend. And he would be doing a lot of reading of scans that would come in. And it was blown away because I asked him, so how many scans do you do in a day? And I was thinking it was a few dozen. And he's no, it's like 1500 to 2000. And I said, how in the world do you do that? And he said, well, we have a team in India who are doctors who are reviewing these before they come to me. And so they're highlighting on a chart what I should look at. But to me, it's a scenario that you could potentially replace that doctor in India with AI and have AI do these screenings that a human was once doing. But in both cases, he was telling me that even if it was a physician in India looking at a chart, he would still look at that scan himself to make sure they didn't miss anything. And that's my big question. I can see this becoming a category shift, but I also see it being really complex because AI could miss a diagnosis of a potential tumor in one of these scans. So I got bad news for you. So could he. So could he. As I quote Biden in the book, it's often said, don't compare me to the almighty, compare me to the alternative. And I think when we look at these AI versus humans versus AI plus humans scenarios, we have to start with the recognition that humans are not perfect. And we get it wrong a fair amount. And so if our standard is the AI has to be perfect, then I think we're going to miss out on a lot of opportunity. I think it doesn't have to be perfect to be better. It just has to be better. And I guess my assumption is again, I'm influenced a little bit by the fact that I take a driverless car once a week, a thing incredibly high stakes. If you've ever been to San Francisco, making a left turn across the Visadero Street, I can't imagine something more complex than that. And so far in many millions of miles driven, there's not been a fatal accident involving a Waymo. And I think that it's just a matter of time. And it's already happening that AI will do at least do that first read of that scan and replace that doctor in India and probably do it at a lower cost than the doctor in India. It raises a whole different set of questions, which is the propping up the middle class in many other poor countries has been through efforts like these that may change. But yeah, I do think that for the foreseeable future, we're going to need what's often called the human in the loop. In this case, the doctor or the radiologist in the loop to do the final reading, the final oversight, because the AI is not going to be perfect. And we're not going to completely trust it. So to me, in some ways, a fundamental challenge for this stage is the AI is going to be good enough to be useful, but not perfect. And therefore the doctors are going to need to look over at shoulders. I think that's probably going to be a better system, a more efficient system, a system that can deliver high quality curate readings at a lower cost than what we have now. There are some flaws in it. And I'm guessing if your friend, the radiologist gets 100 scans in a row where the AI has been perfect, the idea that he's looking really carefully at scan 101, just as opposed to signing off on it. He's a human being. He probably is not. There's a second problem, which is that we're now beginning to see studies where the AI is so good that when the human then weighs in, they actually make things worse. They actually degrade the performance of the AI. And so we've got to be sensitive and attuned to that. And the third problem is what's known as de-skilling, which is even if your friend stays alert for scan 101, over time he may be less good at it than he was in the beginning. And there was just a study that came out last year of gastroenterologists looking at doing colonoscopies. And for three months, they had access to AI-enabled colonoscopy. It puts a little box around the lesions inside your colon, and they liked it and it improved their performance. Then they turned it off. And their performance for the subsequent three to six months was lower than it had been before they used in the first place. And these were not newbies. These were people who'd been doing this procedure for an average of 10 years. So this, I think it's going to be a huge number of interesting, complex questions of how do we get this dyad of AI plus humans right? But I think that your friend, the radiologist, is going to eventually feel like this is actually helping him do his job, able to do more scans and do them better. The issue, of course, will be at some point, when does he get ditched? And the average US radiologist makes about half a million dollars a year. If the AI can read the scans as well as he can, but it costs $50,000 a year to bring into a system, he probably is, he's probably going to be out of a job. I don't think that's happening anytime soon, but hard to believe it doesn't happen within 10 or 20 years. I hope. Now, the next story I wanted to go to is when we think about some of the major social media networks, bias and misinformation have become huge issues. And when I think about medicine, and it's something you touch on in the book, medicine is uniquely vulnerable to AI overconfidence and to biases and that misinformation. So what makes those things more dangerous in healthcare than in almost any other field? Well, the stakes are incredibly high. The emotions are high in healthcare. Our costs are unbelievably high. There are very few other fields where the frontline practitioner, let's say you're a lawyer or an accountant or a businessman, I could easily through the stroke of a pen or a keystroke prescribe a medicine or order a procedure that will cost $100,000. So the stakes are high, both in life and death and in money. I think the bias issue is we're particularly vulnerable in medicine in the corporate world. You're selling to people that are paying for whatever you're selling out of their pocket by and large. And if you're a company, you probably have no aspiration to sell to all comers. There's a population who is eligible to buy the thing that you're selling. And those are the people that you're trying to cater to. You walk into my ER, you will see millionaires and you will see homeless people. You will see people who are English as their fourth language and the AI tools that we might build for a healthy 30-year-old have to also work for a 80-year-old person who speaks only Mandarin and lives in a third floor Chinatown walkup. That's a pretty hard thing to do. The biases are also very real. There have been many studies in healthcare that show that if you take the same patient who comes in the ER with a fracture, if the patient is white, they get more pain medicine and the patient is black. And that means humans are biased. But if AI takes what it sees in the medical record and says that is the right way to deliver care, it basically encodes the biases in its software and therefore, it scales it. And I think the real issue of a lot of the things in medicine are we can hurt somebody one patient at a time if the AI gets it wrong. Let's say a psychiatrist who's really bad can maybe hurt 100 patients or 1,000 patients. A chatbot doing mental health that's really bad could hurt a million patients. So the scaling, I think, is really a problematic issue. But again, compare me against the, not to the almighty, these are human problems. Now that we have the capacity to scale it, we've got to think about them in a very different way. I've had Professor Ethan Mollack from your alma mater, Pennsylvania on the show. And you quote him in the book, the AI you're using today is the worst AI you'll ever use. Do you find that helpful or terrifying in medicine? In medicine, more hopeful than terrifying. First of all, I think Ethan's one of my heroes. I think he is really more one of the most thoughtful people and what AI is going to do to every industry. Yeah, again, I think the stakes of medicine are so high. The work is so hard that having it become better and better to me has to be a net positive. And now where is it potentially terrifying? I like my job. If it becomes so good that it replaces a doctor, that may be good for patients. It may be that patients can get more accessible, more convenient medical care at a lower cost, because I'm probably more costly than the bot is. So it could be harmful for different sectors of society if it begins taking over more jobs. But if you look at it from the perspective of what is the healthcare system here to do, which is not really to provide full employment, it's to take care of people and do it at a cost that society can afford. That I'm hopeful about. Where I get scared is things like deep fakes. The idea that anybody can be made to look like they're saying anything, I think creates a level of potential distrust, which I think is going to play out in everything in politics and in every other walk of life. And it'll have its own particular flavor in medicine. And I think the American public is with me, the Gallup survey last year asked people how they feel about AI. And they were very negative about what its impact is going to be on politics, unemployment. The only area they were positive about was in healthcare. And I think they're right. So I'm looking at this through the lens of healthcare, whereas the AI gets better, I think it's better and better for healthcare. I think for the rest of society, I'm actually quite fearful of it. It makes it easier to design a bio weapon, for example. Yeah. So as I mentioned earlier, one of the things I really concentrate on is human flourishing. And I have a lot of friends who are in the medical profession. And I've probably not seen a group of friends that I have that are more burned out than doctors. And so many of them feel indifferent. And it's almost like they're losing their humanity because of the way that the system is working right now. Is it fair to say that AI could be the first way that more of those doctors could get their humanity back? I think so. I think you have to disentangle a few different issues that are going into their burnout. The issue of the kind of administrative paper pushing, which is a real issue, I think is very important when we rolled out these AI scribes to create the documentation at UCSF, the feedback from the physicians was ecstatic. People said I was thinking of retiring and I'm not anymore. So I think that's low hanging fruit. I think that the paperwork, the dealing with the insurance companies and back and forth, I think those things are relatively straightforward and I think will help quite a bit. The other parts of this though, I think there's some more fundamental problems that are going into clinician burnout that I'm not sure AI is the answer to. It used to be that physicians were uniformly trusted and respected, think back to the days of Marcus Welby and Dr. Kildare. There's a lot of skepticism of expertise in general and I think this played out during COVID. I think physicians feel that. Physicians went into this field because they want to help people and having patients come in being quite skeptical that what you're talking about or pushing back on evidence-based recommendations is frustrating and you get tired of having those arguments. I don't know that AI helps that and in fact with the world of deep fakes combined with social media, I could actually see that getting worse. So there are parts of this that I think make the job of being a doctor or a nurse better. A lot of it is the paperwork part but the issue of trust and there's also some built-in conflict in the medical system that I don't know AI is going to completely fix and that is because of the third-party payers. Again, when you're going to buy something from the store, it's you and the store and you make a decision whether it's worth it. Here, as we get better and better therapies for obesity or for Alzheimer's or for cancer, but these things cost tens and sometimes hundreds of thousands of dollars a year, there is going to be an inevitable push-pull between patients who want a thing and insurance companies or payers who don't want to give you that thing and physicians find themselves often right in the middle of that and it's a really uncomfortable place to be. I don't know that AI fixes that. I want to talk about the other side of this. So there's the physician side and then there's the patient side. So one of the big things I've found in my own life and I advocate to other people is becoming the CEO of your own health journey because no one knows your body better than you. No one can see all the information better than you and I think AI can help people become more empowered patients but it does change in some ways the power dynamic and care. Do you think it does it for better or for worse? I think for better, I'm a big believer in the democratization of care. I think when a patient comes into the office assuming I have time, which is its own issue, and they've done some preliminary research and the research they can do today, putting their information in GPT or Gemini is much better than they could have done three years ago, putting it into Google. I think that's good. I think they're more informed, they're more engaged. I think that's net good. I think where we're going to get in trouble is that the tools right now aren't great for patients and what I mean by that, the Washington Post did a piece a couple of months ago where the digital journalist Jeff Fowler asked me to review 12 interactions between patients and GPT on healthcare. So they had questions about their health. A lot of the things that GPT said were terrific and one of them actually was far better than what I would have said. It understood some new nuance in the literature that I wasn't aware of, but some of them were awful and one of them was shockingly bad and I actually potentially, I thought it even fatal. When I put stuff into GPT or Gemini and get an answer back, I'm an expert. I can look, first of all, I know what to put in and second of all, I see the answer and I say, that's actually quite smart and that other thing it's telling me is ridiculous and I'm not going to listen to it. Patients have no ability to do that. I watch my daughter go through med school. That process is basically going from late person to expert. And so the tools that patients have to use AI to help them essentially do self-care, I think are somewhat limited and I think they'll get better. I think I can envision a tool where it's not you just put all your history into GPT, but it's you say, GPT, I woke up this morning and I'm dizzy or I'm short of breath and then the tool says to you, well, tell me, here are a bunch of questions I need to have the answers to in order to understand it. Do you have a fever? Are you having chest pain? And you have a back and forth discussion as you would with a doctor. I think that would make patient facing AI much more useful using generic tools like Gemini and GPT, I think are pretty imperfect for that function right now. One of the comparisons that you made in the book is that this moment in time is similar to photography replacing portrait painters. When technical mastery fades, what remains essential to being a doctor? I guess we'll see because I think we used to think empathy and yet these tools are capable of faking empathy pretty darn well. Which for those of us who are hanging on to your question, like I went to 10 years of school, what is it that's unique about me? I think it's a tricky question. I would not want a bot to tell me that I had a really dreaded disease. I think that that's going to be really important. Now, is that forever? I don't know, but for the foreseeable future, if I have a new diagnosis of kidney failure or diabetes or cancer or AIDS or you name it, I would want to know that from a human and I want that interaction. I wouldn't want the AI to make a high stakes decision for me, the decision to start chemotherapy or take me to the OR for surgery. I want a human involved in that. For many patients, simple advice a little bit, but I think correctly, the care of your blood pressure, your care of your cholesterol, the care of your diabetes, once we know you have the diagnosis, I think actually could be done by AI and done well, safely, and more conveniently and probably less expensively. But a lot of patients have five things wrong with them. I wouldn't trust AI to get it right because when you have five things wrong with you and you have a new symptom, because there are a lot of interactions between those things that are hard to parse and I think humans are better. I think a lot of patients need a lot of fairly complex team care. I think humans are better at coordinating that and interacting with social worker pharmacists to try to get you the care you need. So I think there's a fair amount left for doctors to do and I think there will be for the foreseeable future, which is partly why I don't think they're going to be unemployed doctors in my lifetime, but I do think doctors are going to be able to augment what they are able to do, probably see more patients, manage more patients in a way that still is more satisfying and better for the patients. I have two last more fun and philosophical questions for you. You've spoken beautifully about how your parents have given you freedom to define success. How has that shaped how you have parented and how you mentor younger physicians today? Parented? I think I've been really quite open to my kids doing whatever it felt right for them and one of my kids is Miley Autistic and what is success for him is a very different version of success than for my other son or daughter. One of my sons works in Major League Baseball, that is Baseball Analytics, another is a doctor. My younger son works in an escape room downtown in San Francisco and success to him is his ability to greet you when you come in and be the voice of God when you're unable to get out in this puzzle. He's never going to be the president of anything or the CEO of anything, but he's wildly successful because he's really achieving his full potential and doing it with a fair amount of joy. Being open to all sorts of ways of we only have this life and lots of ways of defining success and a lot of it relates to are you doing something important, making a difference in people's lives and are you enjoying yourself? Are you feeling that you have good friends and family? With mentees, I think a lot of it is the same thing, trying to come into a mentoring relationship without a mold of what success looks like or being too egotistical and success looks like. Look at me and how well I've done that should be your model. I've mentored a lot of people to do a whole variety of things over the years and it's one of the more gratifying things I get to do to see that their success might be working in AI outside of medicine these days and it might be being a primary care doctor and it might be being a cardiologist and there are all sorts of ways of spelling success. Awesome and then Bob, last question for you is you're on the podcast Passion Struck. What does it mean to you to live a Passion Struck life? Well, I love your emphasis on flourishing. I think that that's a word that I don't think people use very much until the last several years and I think you've partly brought it to public attention. To me that is I'm now 68 years old. I try to look at every day, am I making a difference in people's lives? Am I being a good husband, a good father, a good grandfather? Am I doing the things I need to do and work but also still playing the piano and playing golf and doing the things that give me joy and on a good day I feel like I've gotten that balance right but every day you've got to recalibrate because the world changes around you. Thank you for that and Bob, thank you so much for being here. Where's the best place that our listeners can go to learn more about you? Well, I have a sub-stack now where I'm mostly talking about AI and healthcare. I have websites and things where they can find me. I'm on Twitter all the not very much these days and my book A Giant Leap will be out I think when we post this so I hope people have a chance to take a look if they're interested in the topic. Awesome. Well, Bob, thank you so much for joining us today. It was an honor to have you. Thank you, John. It was real pleasure. That brings us to the end of today's conversation with Dr. Bob Wachter. What struck me most is how nuanced this moment really is. AI is not simply replacing humans, it's reshaping the roles humans play. In medicine that could mean less time spent on paperwork and bureaucracy and more time focused on empathy, judgment, and human connection. But it also raises profound questions such as who do we trust? How do we guard against bias and misinformation and what does care look like when machines become part of the decision-making process? Dr. Wachter reminds us that technology alone doesn't determine outcomes. Human values oversight and intention still matter enormously and that leads directly into our next conversation. On Thursday, I'm joined by Bianca D'Alessio, entrepreneur real estate leader and founder of one of the fastest growing luxury sales teams in the country. While today's episode explored how systems are changing from the outside in, Bianca's story focuses on transformation from the inside out. We discuss ambition, identity, leadership, and what it takes to redefine success on your own terms. It's a powerful continuation of the life beyond the script series. I think the most important thing is realizing that we all only have one life to live and we have to live it for ourselves. And you will exhaust yourself so much trying to live life for everyone else. And that's so much easier said than done because I've been on that hamster real many times. But I think that's why for me, a constant recalibration of understanding like, what are my goals? Why am I doing this? Is this serving me? Is this relationship serving me and helping me become the best person I want to be? If this episode resonated with you, share it with someone who cares about the future of healthcare. Leave a five star rating review on Apple Podcast or Spotify and explore more reflections from this series at theignitedlife.net. Until next time, remember, the future isn't something that simply arrives. It's something we shape through the choices we make, the systems we build, and the values we refuse to abandon. I'm John Miles, and you've been Passionstruck.