# Charlotte Blease: Is A.I. Going to Remedy Medicine’s Problems?

An episode of [Ground Truths](https://invisiblenetwork.ai/podcast/1728526108), 12 Sep 2025, 31 min.

Audio: https://api.substack.com/feed/podcast/172578327/229201fe212178a21b5b7f4941a7fa33.mp3

Thank you Bruce Lanphear, Clyde Wilson, Tracy Dennis-Tiwary, Diego Pereyra, Dr Mike Hunter, and many others for tuning into my live video with Charlotte Blease! Join me for my next live video in the app. Whether A.I. will transform the practice of medicine in a positive way remains controversial. Health researcher Prof Charlotte Blease, on faculty at Uppsala University in Sweden and researcher at Harvard University, has written a new book —Dr. Bot—that critically assesses the unmet needs in healthcare and whether A.I. can fulfill them. She provides an optimistic viewpoint (see the subtitle), and our conversation probes whether that is justified. She has a substack, too, here Thanks for listening to Ground Truths. Through analytic essays and podcasts, I try to cover the important issues and discoveries in life science and medicine. If you have suggestions for topics I should get into, please pass them along. Get full access to Ground Truths at erictopol.substack.com/subscribe

## Transcript

[0:00] SPEAKER_00: Now, this is very provocative, Charlotte.
[0:03] SPEAKER_00: And to start off, there's a lot of common ground that we've covered in
[0:08] SPEAKER_00: like my book, Deep Medicine and your new book.
[0:11] SPEAKER_00: And we're also at a time when AI is getting lots of grief,
[0:14] SPEAKER_00: lots of pushback,
[0:16] SPEAKER_00: blowback,
[0:17] SPEAKER_00: everything.
[0:18] And now you've come out with a book that,
[0:19] you know,
[0:20] this sounds like AI is their cure.
[0:23] SPEAKER_00: So let's start off,
[0:25] SPEAKER_00: if you would,
[0:26] SPEAKER_00: you come up with a really,
[0:27] SPEAKER_00: some great vignettes of patient experience.
[0:31] SPEAKER_00: And one is Jen Lawson.
[0:35] SPEAKER_00: Um,
[0:35] SPEAKER_00: a young woman with Ehlers-Danlos syndrome.
[0:38] Why don't you take us through that to kind of exemplify where
[0:42] the problems are in
[0:43] SPEAKER_00: medicine today?
[0:43] SPEAKER_01: Yeah,
[0:45] SPEAKER_01: and I would say,
[0:46] SPEAKER_01: Eric,
[0:47] I don't want to fall into this kind of false dichotomy of
[0:51] either-or camps here.
[0:53] SPEAKER_01: I mean,
[0:54] SPEAKER_01: the book is a very provocative title in some sense,
[0:57] but basically what I'm trying to do is say what are the
[1:00] problems with current
[1:03] SPEAKER_01: disease
[1:04] traditional medicine to which AI could potentially be a
[1:08] solution or might not,
[1:09] SPEAKER_01: thinking about the status of current AI.
[1:12] SPEAKER_01: But yeah,
[1:13] SPEAKER_01: Jen Lawson is a patient with Ehlers-Danlos syndrome who,
[1:18] SPEAKER_01: through a series of problems with her health,
[1:22] SPEAKER_01: loses her medical and she has a very serious fall.
[1:27] SPEAKER_01: She loses her health coverage and basically eats into her pension fund.
[1:32] SPEAKER_01: in order to cover her health costs.
[1:35] SPEAKER_01: And as we know, that's not a very uncommon story.
[1:38] So what I do is I open it up with her to discuss not only the
[1:43] financial barriers to
[1:44] SPEAKER_01: care,
[1:46] SPEAKER_01: which sometimes even get hidden when it comes to thinking about universal
[1:52] SPEAKER_01: healthcare,
[1:53] SPEAKER_01: healthcare-free at the point of access,
[1:55] SPEAKER_01: so to speak, is going to be a solution to some of medicine's woes.
[2:01] SPEAKER_01: It will be for some patients.
[2:03] SPEAKER_01: But what I'm trying to say is there are multiple barriers,
[2:07] SPEAKER_01: even given the fact that,
[2:09] SPEAKER_01: say,
[2:09] she's able to get access to care because she has to rely on her
[2:13] elderly mother to
[2:14] SPEAKER_01: drive her across state to see a specialist at one point.
[2:18] SPEAKER_01: it's very uncomfortable.
[2:20] She has to pay out of pocket in order to do that because
[2:23] realistically she can't
[2:24] SPEAKER_01: rely on the,
[2:25] SPEAKER_01: on the,
[2:26] SPEAKER_01: the budget options that would be covered.
[2:30] And she's basically got a whole series of obstacles that
[2:34] prevent her from having
[2:36] SPEAKER_01: comfortable access to care.
[2:37] And then I say,
[2:38] you know,
[2:39] with telemedicine,
[2:41] some of the challenges were eradicated and,
[2:44] um,
[2:45] SPEAKER_01: she was able comfortably to access some doctors,
[2:49] SPEAKER_01: you know,
[2:51] with pillows behind her and all the rest of her where she can
[2:53] sit comfortably.
[2:54] SPEAKER_01: But Jen suffers from sort of,
[2:59] SPEAKER_01: not just sort of the literal barriers to care,
[3:02] SPEAKER_01: but figurative barriers too.
[3:03] She discusses some of the biases that do kick in that make it
[3:07] more difficult to see
[3:08] SPEAKER_01: patients with extra needs and to make them comfortable
[3:12] And I use that as a vehicle and as a way to discuss the
[3:18] challenges for doctors and
[3:20] SPEAKER_01: nurses and other people to multitask and to do things to make.
[3:24] But ultimately,
[3:25] she exemplifies,
[3:26] to cut a long story short,
[3:28] the inverse care law.
[3:29] SPEAKER_01: That is,
[3:31] SPEAKER_01: for the people most in need,
[3:33] there are more barriers to getting access to bricks and mortar
[3:36] health care.
[3:38] SPEAKER_00: Yeah,
[3:38] SPEAKER_00: no,
[3:38] I think it exemplifies the problem where her 80-year-old mother
[3:43] had to drive her
[3:44] SPEAKER_00: two and a half hours to get to the doctor.
[3:47] SPEAKER_00: And there was very little time at which she could have.
[3:49] SPEAKER_00: And you get to the point about where medicine,
[3:52] SPEAKER_00: you know,
[3:53] SPEAKER_00: 250,000 people die each year in the United States from errors.
[3:58] SPEAKER_00: 800,000 from disability total or deaths.
[4:02] And so this issue that we have today in medicine,
[4:06] summarized by that doctors aren't superheroes.
[4:11] SPEAKER_00: Generally,
[4:13] SPEAKER_00: we have a lot of deference,
[4:15] SPEAKER_00: the public to doctors,
[4:17] SPEAKER_00: and patients are often self-centered.
[4:20] SPEAKER_00: self-censored about things that they would share.
[4:24] And there's only 18 seconds before they're interrupted to tell
[4:27] their story and all
[4:27] SPEAKER_00: those sorts of things.
[4:28] SPEAKER_00: So that's the backdrop to what an AI do.
[4:33] And let's get into that now because you do like a side-by-side
[4:38] comparison.
[4:40] of medicine today versus AI,
[4:44] whereby,
[4:45] for example,
[4:46] doctors get tired,
[4:47] AI is indefatigable.
[4:49] SPEAKER_00: Take us through that by comparison, if you will.
[4:52] Yeah,
[4:53] so basically the book follows the trajectory of the traditional
[4:59] medical appointment.
[5:01] SPEAKER_01: So from when you disclose your symptoms
[5:06] to a doctor and I consider the challenges with that because
[5:11] sometimes that's
[5:12] overshadowed and then I look at as you say what could
[5:16] technology do as a workaround
[5:19] for that I also consider diagnostics so the challenges on
[5:28] doctors to keep up to
[5:29] SPEAKER_01: date the noise in medicine in terms of the range of
[5:35] SPEAKER_01: if you will, disparities in terms of how a doctor might judge you.
[5:38] SPEAKER_01: It could be gender,
[5:40] SPEAKER_01: it could be age,
[5:41] all of these things could be the time of day that can nudge
[5:45] doctors' decisions in
[5:46] SPEAKER_01: different directions.
[5:49] SPEAKER_01: And I consider what AI can do there as well,
[5:53] SPEAKER_01: not just to keep up to date,
[5:55] SPEAKER_01: but also in terms of eliminating some of that noise.
[5:58] SPEAKER_01: But of course,
[5:59] SPEAKER_01: I'm not giving the technology,
[6:00] SPEAKER_01: it's not a love letter to technology either,
[6:02] SPEAKER_01: because
[6:03] SPEAKER_01: we know that there's challenges there too.
[6:05] SPEAKER_01: So in a sense,
[6:06] SPEAKER_01: probably I'm irritating both sides in this to some extent,
[6:10] SPEAKER_01: but I'm trying to be sympathetic to what doctors have to do.
[6:15] SPEAKER_01: And then I also consider empathy.
[6:17] Empathy is a big one where I think sometimes there's an
[6:20] assumption that look,
[6:21] SPEAKER_01: and I've done a lot of surveys on this,
[6:24] SPEAKER_01: when it comes to bedside manner,
[6:26] SPEAKER_01: that that's a soft skill.
[6:30] SPEAKER_01: And the psychological research shows that it's,
[6:33] SPEAKER_01: It's not particularly easy,
[6:35] SPEAKER_01: but doctors routinely,
[6:36] SPEAKER_01: when I've asked them in surveys,
[6:38] SPEAKER_01: even laterally with generative AI surveys,
[6:42] which task do you think is most likely to be replaced or least
[6:47] likely to be
[6:47] SPEAKER_01: replaced?
[6:48] The one they always say is the least likely to be replaced is
[6:51] empathy,
[6:51] right?
[6:52] SPEAKER_01: Nine in 10, that's a pretty robust figure.
[6:54] SPEAKER_01: A lot of surveys they've done since 2018.
[6:58] SPEAKER_01: So the point is to say, what can AI do?
[7:03] SPEAKER_01: for each of these variety of challenges.
[7:07] And I'm trying to focus more carefully on what the
[7:15] psychological challenges and
[7:16] SPEAKER_01: pitfalls in medicine are.
[7:20] SPEAKER_00: Yeah.
[7:20] SPEAKER_00: No,
[7:20] SPEAKER_00: I think it's really interesting,
[7:22] SPEAKER_00: your approach here,
[7:23] SPEAKER_00: because you basically,
[7:24] SPEAKER_00: you know,
[7:25] SPEAKER_00: review the serious problems we have in the practice of medicine today.
[7:29] SPEAKER_00: And then you get into where AI, you know, could lend a helping hand.
[7:34] SPEAKER_00: And you also review some of the liabilities of AI.
[7:37] SPEAKER_00: So, like you said, it's not a love letter for AI.
[7:40] SPEAKER_00: But there's a lot of people right now who have antibodies to AI.
[7:46] Like,
[7:46] for example,
[7:47] there was a Wall Street Journal essay on the 30th of August.
[7:54] SPEAKER_00: A chatbot can be part of your medical team.
[7:58] SPEAKER_00: And this was by Tom Rosenblatt in the Wall Street Journal.
[8:02] And he was basically giving tips to patients about how they
[8:05] could use chatbots.
[8:06] SPEAKER_00: And as you know,
[8:07] SPEAKER_00: a billion people around the world using chatbots and,
[8:11] SPEAKER_00: of course,
[8:12] are using it to help their health of themselves or their family
[8:15] members or friends.
[8:16] SPEAKER_00: And you actually reviewed some of that in the book, too.
[8:19] But when that was posted,
[8:22] I put it up because I thought it was interesting.
[8:25] SPEAKER_00: There's all these people saying, you know, hard pass.
[8:27] SPEAKER_00: Oh, this is horrible.
[8:28] SPEAKER_00: I don't want an AI.
[8:30] SPEAKER_00: So we have a lot of problems here with negativism,
[8:35] SPEAKER_00: both from patients,
[8:37] SPEAKER_00: prospective patients,
[8:38] SPEAKER_00: and doctors too,
[8:39] SPEAKER_00: by the way.
[8:39] SPEAKER_00: Definitely.
[8:41] SPEAKER_00: Very different way.
[8:44] SPEAKER_00: You're not a doctor.
[8:45] SPEAKER_00: You're not a nurse.
[8:47] SPEAKER_00: Like you said in the book, you don't have any stake in this.
[8:49] SPEAKER_00: You're trying to be an objective, neutral critique observer.
[8:56] SPEAKER_00: Think about all this pushback about AI, and what is the basis of it?
[9:01] SPEAKER_01: Um, so just to say my background is philosophy.
[9:05] SPEAKER_01: My PhD is philosophy of science and mind.
[9:07] SPEAKER_01: I veered into theoretical psychology,
[9:09] SPEAKER_01: which made me interested in telling the,
[9:12] SPEAKER_01: giving a particular portrait of medicine that I felt was missing.
[9:17] Again,
[9:17] for understandable reasons,
[9:18] doctors can't be experts in all things medical.
[9:21] SPEAKER_01: So it's looking at that,
[9:22] SPEAKER_01: that portrait of,
[9:24] SPEAKER_01: uh,
[9:25] SPEAKER_01: the pitfalls,
[9:25] SPEAKER_01: the psychological pitfalls with current medicine.
[9:28] SPEAKER_01: Um,
[9:29] SPEAKER_01: So there is,
[9:30] SPEAKER_01: but then with my philosophical hat on,
[9:33] SPEAKER_01: I'm keen also to take an objective perspective here.
[9:40] SPEAKER_01: And I think there's a big worry current.
[9:44] SPEAKER_01: I mean, social media and so on can be very vitriolic.
[9:48] SPEAKER_01: And I mean, that sort of showcases the worst.
[9:51] SPEAKER_01: And we know that's a problem in itself.
[9:53] But when I'm in conversation with people,
[9:56] I think there's a hunger for the middle ground.
[9:58] SPEAKER_01: I think many people,
[9:59] SPEAKER_01: and there is this sort of,
[10:00] SPEAKER_01: again,
[10:01] SPEAKER_01: this bifurcation between fear,
[10:05] SPEAKER_01: utter fear and existential threat on the one hand,
[10:08] SPEAKER_01: and the kind of fantasy on the other,
[10:10] SPEAKER_01: sort of utopia versus dystopia when it comes to AI and healthcare.
[10:15] I think there's understandable reasons why people get emotional
[10:18] about it and upset
[10:21] SPEAKER_01: on both sides.
[10:22] SPEAKER_01: But I also hear,
[10:24] SPEAKER_01: and again,
[10:24] SPEAKER_01: I've been doing a lot of survey research in this,
[10:27] people recognize that you can take a practical perspective that
[10:31] there are going to
[10:32] SPEAKER_01: be some benefits.
[10:34] And of course,
[10:35] no innovation solves all problems while not inviting new ones.
[10:40] SPEAKER_01: We know this through the history of science.
[10:42] SPEAKER_01: That's just the nature of things.
[10:44] SPEAKER_01: You resolve some problems, you invite many more.
[10:47] SPEAKER_01: But I suppose what I'm trying to say is,
[10:52] SPEAKER_01: take some of the considerations, concerns that we may have.
[10:57] SPEAKER_01: Here's one of them, discrimination or bias in healthcare.
[11:01] SPEAKER_01: How good are we at actually de-biasing doctors?
[11:05] SPEAKER_01: I mean,
[11:05] we do sort of anti-bias training and people should monitor
[11:08] their own stereotyping
[11:10] SPEAKER_01: and the American Medical Association advises that.
[11:12] SPEAKER_01: It's pretty ineffectual.
[11:14] SPEAKER_01: So we have to say if we really care about patient outcomes,
[11:18] and things like face-to-face discrimination interfering with
[11:22] diagnostic accuracy or
[11:23] SPEAKER_01: the treatment of patients.
[11:25] SPEAKER_01: Is there a better workaround to this?
[11:27] SPEAKER_01: And that's where I say,
[11:28] SPEAKER_01: actually,
[11:28] it might be easier to de-bias some of these tools than it is to
[11:32] de-bias people.
[11:33] SPEAKER_01: Or you're putting a hell of a burden on a medical doctor in a,
[11:36] SPEAKER_01: say,
[11:37] SPEAKER_01: Dr.
[11:37] SPEAKER_01: Phil parlance being the best version of themselves.
[11:40] SPEAKER_01: That's very challenging in the frenetic world.
[11:43] SPEAKER_01: We need a realistic, practical conversation about that.
[11:47] SPEAKER_00: Yeah, no, I think you're making some good points there.
[11:50] SPEAKER_00: I do want to go back to empathy in a moment because that's something
[11:55] SPEAKER_00: very much, uh, unanticipated.
[11:56] SPEAKER_00: I certainly didn't anticipate it,
[11:58] SPEAKER_00: but before I do that,
[12:00] SPEAKER_00: um,
[12:00] SPEAKER_00: you know,
[12:01] SPEAKER_00: there's a lot of,
[12:02] SPEAKER_00: uh,
[12:02] SPEAKER_00: what you,
[12:03] SPEAKER_00: what you refer to,
[12:04] SPEAKER_00: of course,
[12:05] SPEAKER_00: uh,
[12:05] SPEAKER_00: Kasparov's law from Kasparov that the,
[12:10] SPEAKER_00: the,
[12:11] SPEAKER_00: as everyone knows,
[12:12] SPEAKER_00: he was the grandmaster chess world champion.
[12:15] And when he lost to the machine,
[12:17] uh,
[12:18] he proclaimed that the best of all worlds is the human and,
[12:22] SPEAKER_00: And in medicine,
[12:25] you may have seen recently that there's been several studies
[12:29] where the AI,
[12:31] SPEAKER_00: and we wrote a phenomenon,
[12:33] SPEAKER_00: and I wrote an op-ed in the New York Times about this,
[12:37] SPEAKER_00: that the AI compared to the doctor with AI.
[12:41] SPEAKER_00: Yeah.
[12:42] SPEAKER_00: I was better.
[12:44] SPEAKER_00: Yes.
[12:44] SPEAKER_00: So that doesn't go along with this hybrid thing.
[12:47] Do you think eventually get to the hybrid model being the best
[12:51] for clinicians?
[12:54] SPEAKER_01: It's such an interesting talking point.
[12:57] So what I think has often happened in medicine is there has
[13:02] been this sort of leitmotif
[13:04] SPEAKER_01: There's no threat here because man and machine will work together.
[13:08] SPEAKER_01: And this is sort of Hallmark-esque,
[13:10] Hallmark movie kind of imagery of the doctor and the AI working
[13:14] well together.
[13:15] SPEAKER_01: Well,
[13:15] SPEAKER_01: for years,
[13:16] SPEAKER_01: there has been AI in the form of clinical decision support,
[13:19] SPEAKER_01: and it hasn't worked out amazingly well.
[13:22] SPEAKER_01: Doctors tend to override AI.
[13:24] because of the so-called last mile problem,
[13:26] how do you actually get these two to integrate?
[13:29] SPEAKER_01: That is the sort of $64,000 question.
[13:33] SPEAKER_01: And it's the psychology of the AI and the doctor.
[13:37] So Kasparov's law is interesting because it was a suggestion
[13:41] when he lost to Deep
[13:43] SPEAKER_01: Blue,
[13:43] SPEAKER_01: the thought was,
[13:44] and then you had this era of the Centop tours in chess matches
[13:49] where you had weaker
[13:51] SPEAKER_01: humans
[13:52] SPEAKER_01: paired with a strong computer and a strong process.
[13:57] SPEAKER_01: They were able to beat, at the time, AI computers.
[14:03] SPEAKER_01: But the lesson for medicine there is it's a weaker human.
[14:09] SPEAKER_01: What's interesting is potentially how then do we train doctors?
[14:13] SPEAKER_01: Because if you have,
[14:15] SPEAKER_01: and again,
[14:15] there's a lot of research that domain experts tend to hold
[14:20] their noses more to
[14:21] SPEAKER_01: algorithmic output.
[14:23] So there's a need here to look at that sort of siloed research
[14:27] on how do humans
[14:29] SPEAKER_01: respond to algorithms?
[14:30] SPEAKER_01: Lay people tend to defer more.
[14:33] SPEAKER_01: Experts tend to,
[14:35] SPEAKER_01: it's called algorithmic appreciation,
[14:37] SPEAKER_01: but experts tend to show more algorithmic aversion.
[14:41] SPEAKER_01: But then what you're saying as well,
[14:46] SPEAKER_01: perhaps we are also now,
[14:48] SPEAKER_01: because certainly in the chess world,
[14:49] SPEAKER_01: if you follow the analogy or the comparison,
[14:52] SPEAKER_01: the era of centaurs is now over.
[14:55] SPEAKER_01: There's no partnership that beats AI.
[14:59] SPEAKER_01: So again, what you see from those studies is
[15:03] SPEAKER_01: the doctor and the AI together.
[15:06] The doctor tends to hurt the accuracy of the AI,
[15:09] whereas if you leave the AI alone,
[15:11] it's better.
[15:11] SPEAKER_01: And then this invites very,
[15:13] SPEAKER_01: I mean,
[15:14] SPEAKER_01: they're taboo questions,
[15:15] SPEAKER_01: but I mean,
[15:17] SPEAKER_01: why are we keeping the...
[15:19] SPEAKER_01: So,
[15:20] SPEAKER_01: but I think we need many more of those studies.
[15:23] It invites questions about what the right partnership is and
[15:26] something that you've
[15:27] SPEAKER_01: been writing about and thinking about for many years,
[15:30] SPEAKER_01: certainly in relation to the NHS and education,
[15:32] SPEAKER_01: What training do we actually need?
[15:34] SPEAKER_01: What's the right training in order to work alongside?
[15:38] Because that's probably what's going to happen,
[15:40] at least in the medium term.
[15:42] SPEAKER_01: It's already happening, Eric.
[15:44] I mean,
[15:44] just to add to that,
[15:45] I've done survey research in the UK among GPs.
[15:49] 2024,
[15:49] I asked 1,000 GPs with my team,
[15:56] are you using any generative AI tools for clinical tasks?
[16:01] SPEAKER_01: 20% said yes.
[16:02] SPEAKER_01: We redid the survey 2025.
[16:04] SPEAKER_01: So January this year, year apart, 25%.
[16:06] SPEAKER_01: Those are commercial tools like ChatCPT.
[16:12] So they're using them,
[16:13] but then as you say,
[16:14] are they using them effectively?
[16:16] SPEAKER_00: These are the un-
[16:18] SPEAKER_00: How grounded are clinicians?
[16:20] SPEAKER_00: How are they willing to embrace?
[16:23] SPEAKER_00: And will they ultimately be superseded for certain tasks by the AI alone?
[16:28] SPEAKER_00: I mean, there's so many uncertainties here.
[16:30] But,
[16:30] you know,
[16:30] one thing we've seen going back to the patient side is we've
[16:33] seen
[16:35] SPEAKER_00: People develop mental health problems from their interactions.
[16:39] SPEAKER_00: We've seen suicides.
[16:40] SPEAKER_00: We've seen some really scary stuff that gets into the heads of people,
[16:46] SPEAKER_00: besides the usual concerns about hallucinations,
[16:50] SPEAKER_00: confabulations,
[16:51] SPEAKER_00: all that.
[16:52] SPEAKER_00: What do you say about that?
[16:53] SPEAKER_01: We do have a slew of articles right now.
[16:59] SPEAKER_01: New York Times has done a lot of these articles during the last week.
[17:04] SPEAKER_01: Look, I think it's a serious problem.
[17:07] SPEAKER_01: It's something very, we should be very concerned about it.
[17:11] SPEAKER_01: But I also,
[17:12] SPEAKER_01: and this may seem,
[17:13] SPEAKER_01: again,
[17:13] SPEAKER_01: my sort of philosophical,
[17:15] SPEAKER_01: let's take a step back a big picture perspective here.
[17:19] SPEAKER_01: While there may well be harms there,
[17:21] we shouldn't overly romance what is currently going on when
[17:28] patients lack access to
[17:30] SPEAKER_01: therapists.
[17:31] because this debate's often more about therapy,
[17:33] in fact,
[17:34] than it is medicine.
[17:36] SPEAKER_01: But also something I've looked at is the evidence base for psychotherapy.
[17:40] SPEAKER_01: Now,
[17:40] SPEAKER_01: consider just for a second,
[17:41] SPEAKER_01: if you park all of the challenges with AI and the potential negatives,
[17:46] SPEAKER_01: especially with young people relying on these tools.
[17:50] SPEAKER_01: But psychotherapy,
[17:54] SPEAKER_01: the late,
[17:54] SPEAKER_01: great Scott Lillienfeld was one of the few sort of iconoclastic clinical
[17:59] SPEAKER_01: psychologists and researchers
[18:01] who estimated back in 2007 that 10% of therapy led to worse
[18:07] outcomes for patients.
[18:08] SPEAKER_01: It tends to be not that much investigated,
[18:11] SPEAKER_01: the harms that can be done in face-to-face therapy with a therapist.
[18:14] SPEAKER_01: So there can be negative outcomes, even if you're seeing a human being.
[18:20] SPEAKER_01: So let's also consider that.
[18:24] Here again,
[18:24] we have this issue of we hold humans,
[18:27] we tend to hold humans to a higher account.
[18:30] SPEAKER_01: to AI.
[18:32] SPEAKER_01: So I think there's much,
[18:35] SPEAKER_01: much more work that needs to be done in saying,
[18:37] SPEAKER_01: again,
[18:39] SPEAKER_01: answering this question of compared to what,
[18:41] SPEAKER_01: who is it useful for?
[18:43] SPEAKER_01: Is it sometimes very helpful for people?
[18:47] Do people also tend to use these tools,
[18:50] be a bit more savvy than we're giving them credit for?
[18:52] SPEAKER_01: Because we've heard these debates before.
[18:55] SPEAKER_01: There's echoes with sort of the Dr. Google stuff.
[18:59] SPEAKER_01: where health professionals came in and sort of said,
[19:02] SPEAKER_01: doctors,
[19:02] SPEAKER_01: patients won't know how to use Google.
[19:05] Studies show that actually people are pretty savvy about it,
[19:09] even so-called digital immigrants.
[19:12] But there's also then the issue,
[19:14] I do think that they are probably causing harm to some people.
[19:21] SPEAKER_01: But I think we've got to consider it in the round.
[19:25] SPEAKER_01: There's much more to think about here.
[19:27] SPEAKER_00: So I think the point you're getting at and you did in the book,
[19:31] SPEAKER_00: Dr.
[19:31] SPEAKER_00: Bott,
[19:31] SPEAKER_00: is that,
[19:32] SPEAKER_00: you know,
[19:32] SPEAKER_00: we have to look at it from the broader benefit versus harm ratio.
[19:37] SPEAKER_00: And now,
[19:38] SPEAKER_00: for example,
[19:39] SPEAKER_00: you just discussed where some people could be hurt,
[19:42] SPEAKER_00: could be overriding benefit.
[19:45] SPEAKER_00: But also,
[19:46] SPEAKER_00: for example,
[19:46] SPEAKER_00: going back to what we talked about earlier with medical errors,
[19:50] SPEAKER_00: there's a lot of human medical errors.
[19:53] SPEAKER_00: And we don't tend to acknowledge that.
[19:55] SPEAKER_00: No.
[19:58] SPEAKER_00: And so, yeah, AI will make errors too, but is it a fraction?
[20:01] SPEAKER_00: Yes.
[20:02] SPEAKER_00: So this comparator is really important.
[20:06] SPEAKER_00: I think it's critical.
[20:10] SPEAKER_00: I think people should keep that in mind.
[20:12] SPEAKER_00: Now, empathy.
[20:14] SPEAKER_00: Yes.
[20:14] This is a surprise,
[20:15] Tau,
[20:16] because when we thought about empathy,
[20:19] machines don't have it.
[20:21] SPEAKER_00: Okay.
[20:21] SPEAKER_00: No way.
[20:22] but they can channel empathy because they're trained by
[20:26] everything that's on the
[20:28] SPEAKER_00: internet and Wikipedia and every other thing that's been written,
[20:31] SPEAKER_00: hundreds of thousands of books and whatnot.
[20:35] So it turns out that the patient can understand at least the
[20:42] sense that the AI is
[20:43] SPEAKER_00: empathetic about their medical condition.
[20:46] SPEAKER_00: And as you point out, and there's now several studies that show the same,
[20:50] SPEAKER_00: is that when rated by doctors,
[20:52] SPEAKER_00: when rated by patients,
[20:54] SPEAKER_00: the empathy here is much more impressive from the AI than by the doctors.
[21:00] SPEAKER_00: Now, that doesn't fit with our model in a way.
[21:04] SPEAKER_00: And, of course, is it real?
[21:09] SPEAKER_00: What do you think is the explanation for this repetitive observation?
[21:16] SPEAKER_01: So it's just such an interest.
[21:18] SPEAKER_01: I'm glad that we're talking about empathy.
[21:20] SPEAKER_01: So a few thoughts that I have on this.
[21:25] SPEAKER_01: First of all, empathy is, as I say, it's a very hard thing to get right.
[21:30] SPEAKER_01: And empathy isn't all it's cracked up to be either.
[21:34] So psychologists and neuroscientists and philosophers have
[21:37] actually written about
[21:38] SPEAKER_01: this for a number of years.
[21:40] SPEAKER_01: Empathy tends to be biased.
[21:42] SPEAKER_01: So we tend to be more empathetic to people who are like us.
[21:44] There's also actual physical constraints on the number of
[21:47] people we can be
[21:48] genuinely empathetic to sort of in the kind of bells and
[21:53] whistles empathy where we
[21:55] SPEAKER_01: feel what they're feeling,
[21:56] SPEAKER_01: we understand what's going on in their lives and all the rest of it.
[22:00] SPEAKER_01: So the key issue, I'd say, Eric, is what do we mean by empathy?
[22:04] SPEAKER_01: Because we can't measure what we haven't first defined.
[22:07] SPEAKER_01: And empathy tends to mean a couple of things.
[22:09] It can be that kind of catching the emotions of someone,
[22:12] so emotional empathy,
[22:14] SPEAKER_01: Now,
[22:14] SPEAKER_01: some medical educators say,
[22:16] SPEAKER_01: Rita Sharon is one of them,
[22:18] SPEAKER_01: that you should feel something of what,
[22:21] SPEAKER_01: doctors should feel what patients are feeling.
[22:24] But there are other kinds of empathy and there's what's called
[22:29] cognitive empathy.
[22:30] SPEAKER_01: So that is just grasping the emotional state of the patient.
[22:34] And then there's no point in sort of grasping it if you're not
[22:37] acting in a
[22:38] SPEAKER_01: pro-social or sympathetic way.
[22:40] SPEAKER_01: So it's actually behaving in a particular way.
[22:44] SPEAKER_01: If we assume,
[22:45] even if we sort of park the bias and all the limitations and
[22:49] the burnout that comes
[22:50] SPEAKER_01: with empathy,
[22:50] SPEAKER_01: actually,
[22:51] SPEAKER_01: if you ask patients what they want from doctors,
[22:54] SPEAKER_01: they tend not to want the emotional empathy.
[22:57] SPEAKER_01: I mean, I've done a small study on this.
[22:59] SPEAKER_01: It was an experimental setup.
[23:00] It has various limitations,
[23:02] but there's other survey research to support this.
[23:06] SPEAKER_01: But again,
[23:07] it's very restricted because people don't often interrogate
[23:09] this idea of empathy in
[23:10] SPEAKER_01: medicine.
[23:11] But they don't want their doctor to be a sort of fellow
[23:15] tragedian in their
[23:16] SPEAKER_01: suffering,
[23:17] SPEAKER_01: weeping with them or feeling.
[23:19] SPEAKER_01: They want them to be professional.
[23:22] SPEAKER_01: So they want cognitive empathy,
[23:24] SPEAKER_01: grasping what you're going through,
[23:25] SPEAKER_01: but also showing compassion for you,
[23:29] SPEAKER_01: behaving a particular way.
[23:31] SPEAKER_01: But that enters the door then to can AI do those things?
[23:35] SPEAKER_01: And it doesn't have to feel anything,
[23:36] SPEAKER_01: but it can,
[23:37] SPEAKER_01: and the studies are showing it can identify emotional state,
[23:41] and it can actually display a certain amount of what looks like
[23:45] compassion.
[23:45] SPEAKER_01: Now, here's the interesting thing.
[23:47] SPEAKER_01: Those surveys were blinded,
[23:50] where people said the Chachi BT or different tools look to be
[23:55] much more empathetic.
[23:56] But when you actually unblind and you say,
[23:59] well,
[23:59] look,
[23:59] that was it,
[23:59] so AI surpasses humans.
[24:03] SPEAKER_01: and also says fewer inappropriate things as well to patients.
[24:08] SPEAKER_01: When you unblind it, that drops.
[24:12] SPEAKER_01: So there's University of Southern California study,
[24:16] SPEAKER_01: which I see when you do the reveal,
[24:18] SPEAKER_01: it's about the same level.
[24:20] SPEAKER_01: Now,
[24:22] SPEAKER_01: of preference,
[24:23] humans who are a bit more quirky and inconsistent in their
[24:28] empathy versus the more
[24:29] SPEAKER_01: slick AI.
[24:31] SPEAKER_01: But that introduces the issue of, you know, at scale, can we do this?
[24:36] And will people actually get used to AI and they see it as in a
[24:40] particular way?
[24:43] SPEAKER_01: But also, I think there's this, again, the false dichotomy.
[24:45] Does it have to be a multitasking doctor who has to deliver all
[24:49] of these things,
[24:49] SPEAKER_01: including empathy?
[24:50] SPEAKER_01: I don't think anyone goes through,
[24:52] SPEAKER_01: you know,
[24:52] SPEAKER_01: spends a quarter of a million dollars to become an empathizer,
[24:57] SPEAKER_01: you know,
[24:57] SPEAKER_01: go through medical school.
[24:58] SPEAKER_01: So there's a lot of...
[25:01] SPEAKER_01: issues there with who do we want to empathize?
[25:03] SPEAKER_01: Those are big questions.
[25:05] SPEAKER_00: Yeah.
[25:06] SPEAKER_00: No,
[25:06] SPEAKER_00: that doesn't even get into the human touch story about when you're,
[25:11] SPEAKER_00: you know,
[25:11] SPEAKER_00: human to human bond,
[25:13] SPEAKER_00: which is a very different,
[25:15] SPEAKER_00: what you're ever going to get from reading the chatbots output.
[25:18] SPEAKER_00: But,
[25:19] SPEAKER_00: you know,
[25:20] SPEAKER_00: this is a big topic because it's even been proposed,
[25:23] SPEAKER_00: I don't know what you think about it,
[25:25] SPEAKER_00: that doctors coached for empathy by AI.
[25:28] SPEAKER_00: That is,
[25:29] SPEAKER_00: when you go through the conversation and you say,
[25:33] SPEAKER_00: the AI reviews it and say,
[25:34] SPEAKER_00: well,
[25:34] SPEAKER_00: Dr.
[25:35] SPEAKER_00: Jones,
[25:36] SPEAKER_00: why didn't you ask the patient about that?
[25:38] SPEAKER_00: Why did you interrupt the patient so about this?
[25:41] SPEAKER_00: And, you know, basically coaching clinicians to be better humans.
[25:46] SPEAKER_00: Is that realistic or is that just crazy?
[25:50] SPEAKER_01: I think it's,
[25:51] SPEAKER_01: you know,
[25:52] SPEAKER_01: in some sense,
[25:53] SPEAKER_01: large language models are just one giant plagiarism machine.
[25:56] SPEAKER_01: They're standing on the shoulders of previous human beings.
[25:59] SPEAKER_01: So in some sense,
[25:59] SPEAKER_01: it's sort of,
[26:00] SPEAKER_01: by extension,
[26:03] SPEAKER_01: could be viewed as the intention to be compassionate.
[26:06] SPEAKER_01: I mean, that's taking quite a long lens to it.
[26:10] SPEAKER_01: But
[26:12] SPEAKER_01: I discuss that in the book.
[26:13] And yes,
[26:14] there are programs where people are using these tools within
[26:18] training.
[26:18] And I think that whatever is in the benefit of patients and can
[26:23] improve delivery of
[26:26] SPEAKER_01: compassion and put it in those courts of terms,
[26:29] SPEAKER_01: I think it's got to be a good thing.
[26:31] SPEAKER_01: You know,
[26:31] SPEAKER_01: if you consider younger people who are going through medical school,
[26:35] SPEAKER_01: they may not have an awful lot of life experience with
[26:40] death or chronic illness or addiction or all kinds of things to
[26:46] which I say if
[26:48] SPEAKER_01: these tools are useful,
[26:50] and I suspect aside from formalised within medical education
[26:55] training programmes,
[26:55] SPEAKER_01: which I know some are doing,
[26:58] SPEAKER_01: some educators are doing,
[27:00] SPEAKER_01: but from my surveys,
[27:01] SPEAKER_01: I suspect that's actually probably in the era of online record access,
[27:06] SPEAKER_01: I suspect that should achieve use
[27:08] when it comes to writing documentation that doctors know
[27:12] patients will read,
[27:14] SPEAKER_01: writing it in an accessible and an empathetic way.
[27:17] SPEAKER_00: Right.
[27:19] SPEAKER_00: Well,
[27:20] SPEAKER_00: Professor Bleas,
[27:22] SPEAKER_00: I should say Charlotte,
[27:23] SPEAKER_00: the great book,
[27:24] SPEAKER_00: Dr.
[27:24] SPEAKER_00: Blatt,
[27:25] it's a very optimistic view of how AI can lead to
[27:29] transformation of better
[27:31] SPEAKER_00: medicine.
[27:31] I think during our conversation this morning,
[27:34] or this evening for you in Europe,
[27:38] SPEAKER_00: Uh, this, uh, is clear.
[27:39] SPEAKER_00: You have a very balanced, very thoughtful view about this.
[27:43] SPEAKER_00: This is a work in progress.
[27:44] It's still very early in this genre of AI,
[27:48] even large language reasoning models.
[27:50] SPEAKER_00: We're in the earliest stages.
[27:52] SPEAKER_00: So perhaps,
[27:53] SPEAKER_00: you know,
[27:53] SPEAKER_00: we're trying to speculate what are the issues where it can go,
[27:56] SPEAKER_00: but I give you a lot of credit for,
[27:58] SPEAKER_00: you know,
[27:58] SPEAKER_00: putting together a whole book with lots of research,
[28:01] SPEAKER_00: uh,
[28:02] SPEAKER_00: uh,
[28:02] SPEAKER_00: citing lots of other research, and we'll see how it plays out.
[28:06] SPEAKER_00: But obviously,
[28:07] SPEAKER_00: it's highly controversial right now,
[28:09] SPEAKER_00: and there's people,
[28:11] SPEAKER_00: as you say,
[28:12] SPEAKER_00: it's nice to be in the middle,
[28:14] SPEAKER_00: but there's a lot of ends.
[28:16] You know,
[28:16] I tend to be optimistic in the long view,
[28:19] so I think you are as well,
[28:21] right?
[28:22] SPEAKER_01: That's definitely my angle.
[28:24] I mean,
[28:24] I take the view that technology now is the worst it's ever
[28:29] going to be,
[28:30] SPEAKER_01: I certainly don't think we're there yet.
[28:33] SPEAKER_01: But I also close the book by saying,
[28:35] SPEAKER_01: look,
[28:35] I'm considering certain psychological challenges that doctors
[28:39] face in patients with
[28:40] SPEAKER_01: a traditional visit,
[28:41] SPEAKER_01: but there's a much wider conversation.
[28:43] SPEAKER_01: Those aren't the only ethical concerns.
[28:46] SPEAKER_01: It's about who do we want to run our health systems,
[28:49] SPEAKER_01: what happens to privacy,
[28:51] SPEAKER_01: inequality,
[28:53] SPEAKER_01: what's the point in having diagnostic accuracy if you don't have
[28:56] We can't pay for medications or you're not living in sanitary
[29:01] conditions.
[29:01] So there are enormous issues that I think as a society and
[29:05] civic conversations that
[29:07] SPEAKER_01: we need to have,
[29:08] which is why we shouldn't postpone those and why this is a nice
[29:11] opportunity.
[29:12] SPEAKER_01: We have to keep talking about them.
[29:14] SPEAKER_00: Well, thank you so much for joining today.
[29:17] SPEAKER_00: I'm sorry that we had problems initially connected with.
[29:22] SPEAKER_00: has been really, you know, very interesting, very worthwhile.
[29:26] You got your pulse on a lot of the key issues surrounding the
[29:30] integration of AI and
[29:31] SPEAKER_00: the future of medicine.
[29:32] SPEAKER_00: I hope you'll, I know you'll stay on it.
[29:34] SPEAKER_00: And I know this is just your second book, the first one with no effect.
[29:38] SPEAKER_00: And I'm sure there's many more books in store for you.
[29:41] SPEAKER_00: So thanks so much.
[29:42] SPEAKER_01: I really appreciate the opportunity.
[29:47] SPEAKER_01: Thank you, Eric.
[29:47] SPEAKER_01: It's great to meet you.
[29:49] SPEAKER_01: Thank you.
[29:50] SPEAKER_00: And thanks for everyone joining today.
[29:52] SPEAKER_00: And sorry again for the problems that we had in initially connecting.
[29:56] SPEAKER_00: We'll look forward to your comments.
[29:58] And we'll have this as an archive so everyone can see at their
[30:01] convenience.
[30:02] SPEAKER_00: Thanks, everybody.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.