The Lost Aura of the Physician

Bryan Vartabedian

Welcome to Liminal MD. I'm Brian Vardabedian. I had the opportunity recently to sit down with Dr. John Lantos, who's both a bioethicist and a pediatrician, who has written an editorial in JAMA arguing that AI is the beginning of medicine's identity crisis, but it's the culmination of something that's been going on for a couple of hundred years now. Now that the machines are here, they're actually getting better at the parts of medicine that we thought were irreducibly human. His editorial is called The Lost Order of the Physician in the Age of Artificial Intelligence, and it asks a question that I think about a lot. When a profession's core competencies become reproducible at scale by a machine, what exactly are we defending as a profession? As a reminder, if you're not a subscriber to the Liminal MD newsletter, the link is below. Would love it if you subscribed. I hope you enjoy our conversation.

Advertisement voice

This episode is brought to you by Accenture. When your advertising operations fall out of sync, everything else follows. Spotify and Accenture are working together to reinvent the rhythm of ad sales using automation, analytics, and smarter workflows to simplify campaign delivery and access better data across the business. The result? Less time spent on operations, more time connecting brands with the moments and fandoms that matter most. Learn more at Accenture.com slash Spotify.

Bryan Vartabedian

Well, Dr. John Lentos, welcome to Liminal MD. Great to be here. So I was really drawn to this editorial, The Lost Aura of the Physician in the Age of Artificial Intelligence, that was recently published in JAMA. Congratulations. Thank you. So much here for me to talk about, and I think everyone should read it. I'll put the link in the show notes. But I want to unpack some of that with you today, just to see what drove you to write it. It's a real big picture idea of an aura. And you suggest that doctors have historically had an aura. What exactly was the physician's aura and how did you come up with that concept?

Dr. John Lantos

So the idea of aura derives from the work of the German philosopher and social critic Walter Benjamin, who wasn't writing about medicine, but was writing about the work of art in the age of mechanical reproduction. And his argument was, that before photography and film,

Dr. John Lantos

paintings had a special something that he called aura that distinguished an original work of art from a copy, that there was something hard to define but quite special about the original work of art that people felt when they were in the presence of that work of art, and that mechanical reproduction, photographs, movies, etc., completely changed that. Any work of art, any visual work of art would be infinitely copied with an exact replica of the original. And therefore, there was no such thing as an original anymore. So I started thinking about that in relation to what AI is doing to medicine. People now go to chat GPT or clot or whichever large language model they require. They can put in their symptoms. They can have a conversation. And in a lot of studies, they prefer the interaction with the robot to their interactions with human physicians. So it got me thinking that something is happening that's similar to what Benjamin described with regard to painting and photography, that whatever this special thing that people used to feel about talking to a human physician, for some reason, if you believe these studies, they are now getting from a mechanical reproduction. And we're only in the early stages. So that is likely to get better. And the human doctors are likely to lose whatever that aura was that they once had.

Bryan Vartabedian

So is the aura what's perceived by the patient? I know growing up all of my life, going to see the doctor, you know, I always saw this as someone who kind of knew so much and had gone through all this extraordinary training and had this magical, depth of knowledge that was unparalleled. Was that in my mind? Were the docs chuckling at me at that point, knowing that they really didn't know everything? I mean, who came up with the aura? Is it a patient thing or a doctor thing?

Dr. John Lantos

Oh, no, I think you're exactly right. I mean, I think knowledge, with this one caveat, knowledge was part of the aura. Empathy, perhaps, was another part of the aura. And authority may have been a third part. So I think most people, when they would go see their doctor, would feel a kind of sacredness, a kind of specialness, a kind of nervousness that they were in the presence of somebody who had knowledge, who had authority, who really cared about them and could give them life and death information in a way that might change their lives forever. The only place to get that was a human physician. And so I think that was completely real. I think the challenge of AI is that at least in some circumstances and probably in a growing number of circumstances, people are going to be able to get something similar, maybe not exactly the same. Photograph is not a painting, but maybe the same enough or good enough that for many people, the accessibility, the perceived empathy, and the vastly superior knowledge is going to be, is going to make AI preferable.

Bryan Vartabedian

Let's stick on the aura before AI. You know, I really was referring to the asymmetry of knowledge that we kind of had that kind of disappeared 20 years, 25 years ago with the rise of the internet and the democratization of information.

Bryan Vartabedian

So that was, do you think that was what was behind that aura? You know, was that asymmetry of knowledge or was it trust? Was it authority? What do you think was the core of it?

Bryan Vartabedian

Partly knowledge.

Dr. John Lantos

Partly knowledge. partly the way knowledge is presented and communication skills. So I think one big difference between the current era of AI and the large language models and Dr. Google, who in some ways became a source of information for many people about their illness, is with the pre-AI search engines, You couldn't have a conversation. You couldn't clarify. You'd get a data dump, and it would give you links to 10 different articles or something. But you couldn't question. You couldn't ask for the explanation in a different way. You couldn't have an interaction that mimicked a human interaction. And now you can.

Bryan Vartabedian

Yeah, it was both remarkable. I came out of fellowship. It went into practice just as patients were starting to have access to information. So I cut my teeth in the era where we had that total asymmetry, where we were the only source of information for patients, which is just remarkable to think about, and saw that age where patients came in with printouts and challenged me. But then, you know, I think patients ultimately came to some – of where information fits with the relationship with me. What's so interesting about large language models, when I look at what patients come in with, it's not just information, but they're learning, or they're able to see the process of thought or to see some of the logic behind diagnostics, but they still don't have the depth to kind of process it. So they can kind of ape some of these things, but they still... They're not fully there yet, right?

Dr. John Lantos

I think that depends on the situation and it's changing rapidly. So, I mean, one of the points of the essay that I wrote in JAMA was that AI is the culmination of a process that's been going on in medicine for centuries. which is to abstract the disease from the patient. So the more we can base medical judgment and, you know, the aura that we had on objective scientific knowledge and less on something mystical, magical about being in the presence of a true healer or authority that may be undeserved. nevertheless present. And the more we can say, here are the markers of disease, here are the prognostic factors that tell you your likely chance of survival, and whether you're the doctor today or I'm the doctor tomorrow doesn't matter. We both draw our authority from the same body of knowledge. And the better that body of knowledge is, the less our personal characteristics matter.

Bryan Vartabedian

Once you do that, you've created... unpack that progressive objectification, if we could. That was a big piece of the of the essay. And you kind of talk about this negotiation between technological efficiency and human presence. And you work through some interesting examples, anesthesia, the rise of evidence based medicine, EHR. Can you walk us through starting maybe with Foucault or anesthesia?

Dr. John Lantos

Sure. I mean, I think Foucault's work, which is not about his medical research, but is a history of the development of this objective approach to medical knowledge, which he pinpoints as happening in France around the turn of the 18th century, late 1700s, early 1800s, along with the Enlightenment, the scientific revolution, and the idea that

Dr. John Lantos

medicine will be improved if we can find objective markers. And some of the markers that people started to use in that era were sort of stethoscope and auscultation. So it objectified pulmonary findings of lab work and lab values, blood counts, that sort of thing. And the more of these developed, the more the ideal became to find objective measures of illness so that Doctors wouldn't necessarily have to talk to their patients anymore. And a lot of these early pioneers of this approach said, you know, the subjectivity of doctors and patients is an impediment to effective medicine. In the paper, then, I go through sort of what this led to, which was the rise of, first, scientific medicine through technology. That became centralized in the hospital. I talk a little bit about the work of Paul Starr and the social transformation of American medicine. Francis Peabody writing 100 years ago in an exhortation to Harvard medical students said, once you put your patient in the hospital, they lose all individuality. Hospitals are like industrial machines to generate data. I mean, this was all happening long before AI

Dr. John Lantos

was even, something anybody besides science fiction writers thought about. It led then to sort of a style, a way of philosophy, a way of practicing medicine where everything became quantifiable. And that then was the groundwork for evidence-based medicine, for things like RVUs as a way of measuring the work that doctors do. You know, every little bit of medical piecework became sort of an important objective measure of whether you were doing something valuable or not. That sort of created a world in which computers are obviously going to be better than people because they are better at memorizing facts and applying quantitative objective guidelines.

Bryan Vartabedian

John, let me pull a quote from your wonderful essay. By the early 21st century, clinical reasoning was encoded in protocols, decision rules, and electronic prompts. Professional judgment shifted from originating decisions to managing exceptions, adjudicating among machine-generated outputs while operating under institutional pressure not to deviate. Professional autonomy steadily eroded, and individualized clinical judgment and discretion became a source of risk. You're describing really the industrialization of medicine, right?

Dr. John Lantos

Yeah, that's correct. I mean, you talked about what life was like before Google when you started your training. I mean... Could you have imagined in those days an insurance company telling you you couldn't order a test or a hospitalization utilization reviewer coming by and saying your patient needs to be discharged? I mean, doctors, that didn't happen in those days because doctors were thought to be the only people who had the knowledge, the skills, and the authority to make those kinds of judgments. That era is gone.

Bryan Vartabedian

And that specific example is really offset by evidence-based medicine protocols kind of taking over how patients are cared for and run through the system, right?

Dr. John Lantos

Yeah. I mean, we can be nostalgic for those days. Certainly as doctors, we can see them as the golden age of medicine where we were respected more than we are now, empowered more than we are now. And had the freedom to exercise a wider range of clinical judgment than we do now. At the same time, in that era, there was a lot of non-evidence-based practice variation that was most likely harmful for patients but was never measured or quantified because we didn't have the clinical epi tools or the big data sources to analyze what was working and what wasn't. So I think objectively measured outcomes are better today for most diseases. But the price of that has been this loss of professional authority and the idea that, for the most part, doctors are almost interchangeable. I mean, the rise of hospitalist medicine is another example of this. Forty, fifty years ago, no internist would have sent their patient to the hospital and not been there. inpatient physician. They thought the knowledge they had of the patient's past medical history and life circumstances were crucial to

Dr. John Lantos

high-quality patient care. Now, one hospitalist is on today, another one is on tomorrow. They're both applying the evidence-based guidelines, and patients say it's dehumanizing. They don't know their doctor's name. They have no idea who's in charge, but the outcomes, at least on some perhaps superficial level of how we measure outcomes, bear out that this style of medicine works pretty well.

Bryan Vartabedian

You're talking about the commodification of the physician, right?

Dr. John Lantos

Exactly. And like that, as I say, lays the groundwork. So had today's AI come in 50 years ago, I think it would have been a more uphill struggle for acceptance. But now doctors look increasingly robotic.

Bryan Vartabedian

So you've touched upon things that are sort of good about this industrialization of medicine and, you know, outcomes that are better and the dismissal of sort of rogue ways of treating patients, etc.

Bryan Vartabedian

You're making the point that maybe there's some good things about the loss of aura, right? Maybe there is an upside. Maybe aura was... found it in some level of paternalism and hard asymmetry and a power struggle between patients and doctors. I mean, is that, would you argue that?

Bryan Vartabedian

Yes.

Dr. John Lantos

I mean, I think knowledge asymmetry and power asymmetry are both bad things. I think the more patients know what we know, the more empowered they will be. And like, in my view, probably the better their outcomes will be. But we spend a lot of time studying. We know a lot of stuff. So that power differential will never leave completely. Searching for results on Google or making the New England Journal publicly accessible or writing up the latest medical findings in the New York Times or the Wall Street Journal. levels the playing field a little bit. Changing legal and bioethical norms about patient autonomy levels the playing field a little bit. We're required to disclose information that we didn't used to have to disclose and to empower patients. But I think the challenge, again, picking up on this early essay by Walter Benjamin, is that doctors aren't going to go away. AI

Dr. John Lantos

is going to be an increasingly important presence. So the question is, how are we going to change the way we think about the human role in a world where both doctors and patients have access to sort of the world's literature summarized in easily accessible language, delivered in an empathic way 24-7? Let's talk about the,

Bryan Vartabedian

the future trajectory of AI and a patient who's being diagnosed with maybe a not-so-complicated problem that, you know, where automation from the point of diagnosis almost to the point of decisions about therapy can be done with a large language model. One of the things, if we were to sort of unpack this or go back out of it and say, We have a fully autonomous system. Where are we going to fit in? And I think you kind of hinted this at the very end of the essay. I know you don't have an answer for it, but

Bryan Vartabedian

despite the fact that technology can do it, do you think there's a piece of it that will remain doing despite the fact that technology can do it? Where do you think we'll fit in?

Dr. John Lantos

I think there will always be a role for human healers. I think

Dr. John Lantos

illness, disability, mortality are such fundamentally human experiences that going through them alone or only accompanied by robots would be a horrible way to respond to illness.

Dr. John Lantos

What shifts, I think, with AI, and you alluded to it in your first question, I mean, we used to be the people who knew, and we were the only people who knew, and that was the basis, we doctors, and that was the basis for our aura, our respect, and our authority.

Dr. John Lantos

That, I think, is going to go away. Then the question will be, what's left? Again, to go back to the Benjamin analogy, painting didn't disappear. Once photography replaced the replicability of it or the accuracy of an image depicting a person or a landscape or something like that, painting changed. There are more painters than ever. They just invented new forms like abstract expressionism, cubism, impressionism, which didn't have the same social role as painting had before and didn't have the same aesthetic goals. I mean, the goal wasn't to create a likeness that was as realistic as possible anymore. So I think the challenge for doctors in the age of AI is going to be to think about what it is that we bring uniquely to this uniquely human experience to help patients interact with the new robotic sources of information that are going to empower them to make decisions in a different way. But there will still be decisions to be made, and a doctor's practical wisdom, clinical experience, empathy, and connection, I think, are going to help people find their way through the thicket of information overload.

Bryan Vartabedian

So I guess the question comes, what kind of doctor becomes valuable in a world where competence is really... Almost reproducible, right?

Dr. John Lantos

Yeah. I mean, the

Dr. John Lantos

cartoon image I have been playing with a lot in my head is sort of human doctor, robotic doctor, patient and family in a room together having a conversation where both the doctor and the patient turn to the robot for factual expertise.

Dr. John Lantos

It knows more than you do. But it doesn't have your judgment. It doesn't have your taste. It may be unable to find nuances in the objective information that are important for this patient today with this kind of family and these sets of values. That, I think, the human doctor is going to become sort of the docent, the guide through a new informational process. I

Bryan Vartabedian

had an interesting experience not long ago. I had a child with helicobacter pylori or

Bryan Vartabedian

suspected helicobacter pylori. And for a variety of reasons, the family was interested in – the child couldn't do a breath test. I think it was on a PPI and was inquiring about the use of serology. And I couldn't remember because we've never used serology in the recent – for the diagnosis of HP, and I couldn't quite rationalize to the family why that was the case. So we went on to DocsGPT, which is the Doximity LLM, with the family, and we both read it, and I said, aha, that's the reason why. And the family thought it was so cool. I think there's a piece of... There are my colleagues who say they would never want to do that because they wouldn't want to disclose that they couldn't remember the exact reason why we don't use serology for HP, why we use stool antigen breath test, blah, blah, blah. But the family, I think it was a validating thing for the family. And

Bryan Vartabedian

it empowered me a little bit. So it was kind of an interesting experience. And I think it's not exactly what a lot of doctors think about when they think about their relationship with the family.

Dr. John Lantos

It's a great story. I'll bet most doctors have either had a similar story or fought against, as you say, the temptation to do it out of fear that it would diminish their authority. But my experience has been like yours. When I've done similar things with families, I think my authority rises in their view because not only have I shown humility, acknowledging that I don't know and looking it up. But I've also shown some technological savvy that we can do this right here in the waiting room or in the examining room. We can pull this up, we can have a conversation, and then we can talk about what it says. And families appreciate that.

Bryan Vartabedian

They absolutely do. You know, there's this horrible quote that I hear sometimes, and I'll tell it to you here. Any doctor who could be replaced by AI, deserves to be replaced. And so, I mean, it's a provocative quote because on the one hand, I think we have to participate in our own redefinition, understand what's happening. I mean, what do you think of that? Do you think that's true? When you hear that, what do you think? Are we responsible or where does the responsibility lie for our redefinition? Is it with... Is it with the academy? Is it with the profession as a whole? Or do we as individuals have to find our way? I mean, it's

Dr. John Lantos

a

Bryan Vartabedian

tough question.

Dr. John Lantos

Yeah.

Dr. John Lantos

You know, I think back to some of the pre-AI situations in which technology replaced doctors. I mean, one of my favorites is reading electrocardiograms.

Dr. John Lantos

For a long time, this was a huge challenge for computer scientists. Eventually, they got pretty good. It's still the case that head-to-head with the best cardiologists, the machines are either a tie or slightly worse, although they're much faster, much more accessible, and it's hard to find the best cardiologists in a small rural town or the ER in the middle of the night. So these are... tremendous advance. Did they replace cardiologists? No. Did they bring the expertise that cardiologists could bring to doctors who needed that expertise? Absolutely. Did they improve patient outcomes? Yes. So I think a similar thing is going to happen with AI. I don't think too many doctors are going to be completely replaced, at least in the near future. But I think there will be many situations where

Dr. John Lantos

doctors will be augmented by AI. I think one of the big challenges is going to be for medical education. And medical education has been largely defined by this ideal, again, the one that you started with, that sort of a good doctor is someone who gets sort of you know, encyclopedias worth of biological information crammed into their head for six years. And then they're supposed to instantly recall that and absorb all the new information that's coming up. And that's medical education. I think the whole reconception that AI is going to force will be an earthquake for pre-med and med school and residency training. We're We're going to have to rethink the whole thing. And to me, the most exciting thing about AI is the way it enables us to ask questions like, what is the value of a human doctor when all the information that used to be the source of their authority is now cheaply, easily available to anybody with a phone in their pocket? Similar questions about empathy.

Dr. John Lantos

Chatbots are often rated by patients as more empathic. than human doctors. I don't think what they're doing is what I would call true empathy. But the fact that the perceived level of empathy is higher allows us to start to question, what is this thing, empathy? How does it work? Can a machine do it better than I can? And if so, how did the machine learn to do that? It learned it by reading things that humans have read. I can read those things too. we could learn how to become more empathic by seeing what the machines do.

Bryan Vartabedian

Sherry Turkle, who's a sociologist at MIT, calls this a robotic moment. And it's this point where we lower the bar and we just, we acquiesce to the, you know, the thing that you said before is just good. It's good enough, right? And so there's another... thought leader I heard speaking who was talking about is, is millennial kids who were talking about buying or purchasing something online. And they have the assumption that whenever a human gets involved with an interaction, something has gone terribly wrong. And I've seen that with my own kids too, but it's, it's true. We just, I think it's, we are in this robotic moment where a chatbot is good enough for many circumstances, but not all. And

Dr. John Lantos

Yeah. I mean, Sherry Turkle's one of the most interesting thinkers and writers about this. She's been working on it for 40 years since, you know, MIT started developing devices that could communicate with people and particularly studying children and how they interact with talking machines or talking dolls and how they form deep, meaningful relationships.

Dr. John Lantos

I mean, one of my favorite quotes from her is that we humans are easy marks for something that seems to like us, you

Bryan Vartabedian

know. Yeah.

Bryan Vartabedian

Absolutely. John, you mentioned medical education.

Bryan Vartabedian

And, you know, for me, this is one of the most challenging things going forward for the – to figure out how to shape a curriculum today for even five years from now. We're not even sure what's going to be going on. Eric Topol, who gave our Baylor College of Medicine commencement address in 2016, one of the admissions committee members raised their hand in a faculty meeting before the commencement and said, what should we be selecting for? for our next class of students for the future? What should be the quality? And he said one word, flexibility. But I think he was getting at the fact that when you and I trained, what we learned is what we carried through most of our career, everything from process to documentation, whereas now we're in this constant state of becoming. I mean, we're just, everything is changing so quickly and the mindset of how you educate yourself and how you change your point of view I'm not sure how that equates in medical education because I think it's a complete foundational disruption. What do you think?

Dr. John Lantos

Flexibility for sure. I mean, I remember when we started medical school, the dean got up and said, half of what you learn is going to be obsolete within 10 years. We're just not sure which half. And so even then. Back in the dark ages of the 20th century, there was an understanding that things were changing rapidly, but it was mostly the knowledge base that was changing. Now the social role of the physician is changing too. So

Dr. John Lantos

one of the things that – one of the analogies I look for in this is what's happening in the AI industry itself. There was an interesting podcast. Jack Clark, who's one of the co-founders of Anthropic, was on Ezra Klein's podcast for The New York Times talking about what's going on at Anthropic. And he said, we're no longer hiring coders. That job that used to be the largest category of employees at Anthropic, we don't need them anymore. The machines have completely replaced them. But we're hiring more people than ever, he said. we're hiring people who can look at the output from the machine coding and decide what's really good. So we're looking for people with taste, he said. I think there's an analogy for what medical educators ought to be looking for and training for. That is,

Dr. John Lantos

The focus on facts, the focus on sort of a fixed set of skills that are relevant today and are going to be obsolete tomorrow is no way to go about educating the future workforce. What we need to do is educate people who can sort of look at what's coming, as you did in the examining room, look at what the machine says when a patient asks you a question you don't know the answer to and decide whether it's a good answer. I mean, in your case, it sounds like you got the right answer. You recognize that. You shared that with the family. They respected you. They respected the machine. It was all good. I've done this and gotten an answer I didn't like.

Dr. John Lantos

I will take pictures of rashes and upload them. I'm a general pediatrician. I'm like, I don't know what this is. Let's ask. And it'll say, could be this. And I'll have a conversation with the LLM and it'll say, oh, that's a good idea.

Bryan Vartabedian

That's right. I do worry about trainees who haven't fully developed great clinical judgment from struggling who are using these LLMs to make decisions when they have... They don't have the front work to understand when it may be a good answer or a bad answer. That's my concern.

Dr. John Lantos

Big

Bryan Vartabedian

concern.

Dr. John Lantos

You know, to go back to the anthropic analogy, if you're looking to hire people with taste, how do people develop taste? They develop it by sampling a lot of good and bad data.

Dr. John Lantos

developing a way to figure out what's valuable and what's not. But I could imagine an approach to medical education that would incorporate that hierarchy of values that said, you know, from day one, AI is your partner, but you are ultimately responsible.

Bryan Vartabedian

Yeah. A lot of questions also arising from the... risk mitigation crowd, you know, who's responsible for these decisions and whether we should be documenting when we use an LLM and making a decision. It's almost overwhelming the amount of change that's happening around us. It's exciting too to me personally.

Dr. John Lantos

You know, 25 years ago, I... I was thinking about these issues already and wrote a book called Do We Still Need Doctors? And one of the responses I got was, of course we do. Who else would we sue? You're

Bryan Vartabedian

absolutely right. Dr. Lantos, how do people find out about you or this book that you wrote? And do you have a site or a place where people can find out more about

Dr. John Lantos

you? I have a website, johnlantos.com. Has books, papers, occasional blog posting, and people can track me down there.

Bryan Vartabedian

Yeah, perfect. We need to get you down to do Grand Rounds in our part of the woods. It sounds like you'd give a fascinating Grand Rounds.

Dr. John Lantos

Yeah, I would love to visit. I know I have a bunch of good friends down there on the Miracle Mile.

Bryan Vartabedian

Well, listen, Dr. Landers, this has just been a wonderful discussion about – an amazing editorial in JAMA. I'll put the link in the show notes, but I love this level of thinking. I think this is what needs to happen. I think there's a lot of AI boosterism and everyone's all hyped up about it, but no one's asking the difficult questions. And as we've revealed here, there aren't answers to those questions, but we have to keep asking them because as the technology evolves, we have to... readjust and help guide the patients and help guide young physicians, right?

Dr. John Lantos

Yeah. And thank you so much for inviting me on and probing a little bit. I think these are the sorts of questions that are going to shape medicine for the next generation. And it's up to us old folks to try to get it on the right track. That's right.

Bryan Vartabedian

Well, Dr. Lance, let's have you back again and we'll pick you up after your next editorial.

Inserted grocery advertisement voice

Summer is here. And Ralph's is your destination for hot savings. Find unique items at low prices with our exclusive brands. Fire up the grill with cookout classics like burgers. And don't forget delicious produce like fresh melons. Or treat yourself to summer's sweetest lineup with Kroger Brand's limited edition all-American ice cream collection. Whatever your summer plans, Ralph's makes it easy to enjoy high-quality food at affordable prices. Ralph's, serving SoCal for over 150 years. Thank you.