Subscribe to The Podcast by KevinMD. Watch on YouTube. Catch up on old episodes!

Join psychiatrist J. Wesley Boyd as we delve into the challenges and opportunities AI presents in teaching medicine. From personalized learning to ethical considerations, we uncover how AI is reshaping the future of medical training.

J. Wesley Boyd is a psychiatrist.

He discusses the KevinMD article, “The role of AI in medical education: Embrace it or fear it?”

Our presenting sponsor is Nuance, a Microsoft company.

Do you spend more time on administrative tasks like clinical documentation than you do with patients? You’re not alone. Clinicians report spending up to two hours on administrative tasks for each hour of care provided. Nuance, a Microsoft company, is committed to helping clinicians restore the balance with Dragon Ambient eXperience – or DAX for short. DAX is an AI-powered, voice-enabled solution that helps physicians cut documentation time in half. DAX Copilot combines proven conversational and ambient AI with the most advanced generative AI in a mobile application that integrates directly with your existing workflows. DAX Copilot can be easily enabled within the workflow of the Dragon Medical application to bring the power of ambient technology to more clinicians faster while leveraging the proven and powerful capabilities used by over 550,000 physicians.

Explore DAX Copilot today. Visit https://nuance.com/daxinaction to see a 12-minute DAX Copilot demo. Discover clinical documentation that writes itself and reclaim your work-life balance.

VISIT SPONSOR → https://nuance.com/daxinaction

SUBSCRIBE TO THE PODCAST → https://kevinmd.com/podcast

RECOMMENDED BY KEVINMD → https://kevinmd.com/recommended

GET CME FOR THIS EPISODE → https://earnc.me/ql69jh

Powered by CMEfy.

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast, get CME for this episode by clicking on the CME link in the show notes. Today we welcome back Wes Boyd. He is a psychiatrist, and we’re going to talk about the KevinMD article that he co-wrote with medical student Amelia Mercado. It’s titled “The role of AI in medical education: Embrace it or fear it?” Wes, welcome back to the show.

Wes Boyd: Delighted to be back here, Kevin. Thank you so much for having me.

Kevin Pho: So Wes has been on in the past. Go to KevinMD.com/podcast to search for his name, hear his story and his prior episodes. But today let’s jump right into this most recent one, “The role of AI in medical education: Embrace it or fear it?” Tell us what this one’s about.

Wes Boyd: Well, I got the idea for the article from the medical student I co-wrote it with, Amelia Mercado. It was about a year and a half ago, we were sitting in my office and she was telling me about AI and this thing called ChatGPT, which I had not heard of at the time. And in front of me, it created, as I remember, a 1,500 calorie recipe for dinner tonight. And right in front of me AI generates this.

And so she exposed me to AI, and said she’d been using it extensively in her own medical school studies, and that was the first I’d heard of it. And so over the course of our getting to know one another and talking about the potential benefits of AI, and how she was using it in her own medical education, and once she introduced me to it, how I started using it in all kinds of ways, she’s like, I really want to write this article. So that’s really the genesis of the article. Again, it came about because of my interactions with Amelia and her huge embrace of AI, and then subsequently my own embrace as well.

Kevin Pho: All right. So you are both a medical educator and a bioethicist, so certainly I’m interested in hearing about this issue through that perspective. But just to give some context, from what you’re seeing and observing, in general how are medical students using AI today?

Wes Boyd: They’re using it pretty extensively. And by the way, they’re doing so in some instances with the full support of their medical schools, and in other instances medical schools are sort of turning away and acting as if AI isn’t even there.

But regardless of what the medical schools are doing in terms of embracing or not AI, medical students, in my opinion, are using it pretty extensively. They’re using it, for example, to create study guides, to anticipate questions that might be on exams. And also they’re doing it to sort of personalize their learning. If they know that they have strengths in some areas and perhaps weaknesses in other areas in their own medical education, they’re using AI to both sort that out and then figure out methods for improving where they might have areas of weakness.

Kevin Pho: And in your role as a medical educator, how are you using AI?

Wes Boyd: Since I was first introduced to ChatGPT and really started thinking concretely about the ways I might use it, I’ve used it in all kinds of ways.

One, I have used it to create outlines for lectures I’m giving. A few weeks ago I gave a talk to lab workers about their own mental health and well-being, and used AI to generate areas where lab workers might have particular stresses or difficulties in their job. I used AI to help think about and see the ways in which lab workers might have lots of stress in their own jobs and everyday lives. And also used AI to help me at least create the first draft of solutions for, and guides for, helping lab workers improve their own mental health and well-being, both through things that they could do for themselves but also that their employers could do.

Most recently, for example, I was asked to write an article in Scientific American about ethical issues surrounding the proposal that the FDA take over regulating what are called lab developed tests. And this is an area that I really knew very little about ahead of time. I know a lot about ethics, but I didn’t know so much about lab developed tests, and what FDA regulation may or may not entail, and what the upsides and potential downsides might be. So I actually used AI to generate arguments for the FDA regulation of lab developed tests, and also against the FDA regulation of lab developed tests, and sort of started using that as a framework for me to arrive at my own opinions about the issue.

The other thing, for example, I’ve used AI in a very highly successful manner, is when I’ve been asked to write letters to support my colleagues being promoted, say to assistant, associate, or full professor. And I’m handed a resume that might be 30 pages long, I’m expected to read it and write a letter, hopefully in support of the promotion.

I can actually feed that resume into AI, and provide also a sample letter of a letter supporting one’s promotion academically, and literally in front of my eyes it will spit out, within seconds, a letter that’s not a bad first draft. I usually have to edit them. But so, for example, a letter that might have taken me four or five hours to write, and this is no exaggeration whatsoever, if I’ve got a 40-page CV of someone I do not know personally and I’m being asked to write a letter, I have used AI to generate the letter, print it, sign it, and deliver it within 40 minutes, which is just an astounding timesaver.

Or another example is at the VA where I currently work in Houston, Texas, we had a day-long fair encouraging folks to write their own advance directives. So it was a big push to have as many patients as possible have advance directives. And I was like, oh, it would be great to have a little flyer or a pamphlet from the ethics committee, which I chair. And sure enough, I just put into ChatGPT, please create a pamphlet encouraging the use of advance directives, and within seconds it spit it out, and in that case I didn’t have to edit anything.

So I’ve used it pretty extensively in all kinds of ways, and it’s a tremendous timesaver if you’re using it wisely, for good purposes.

Kevin Pho: Now, among your medical educator colleagues, and I’m sure you don’t discuss who uses AI or not, but in general do you feel that your colleagues are using AI to the extent you are, or even regularly, or is this still a little bit of a foreign concept among your colleagues?

Wes Boyd: It’s a great question. I have no doubt some of my colleagues are using AI way more than I am, but I think that would be a minority. And I think the vast majority are using it far less than I am.

But as recently as last week, for example, I told one of my colleagues at Harvard about the way I had used AI to create letters. And after I had informed her about how I use AI to generate letters, she actually wrote a letter to a potential applicant in our program, and she cc’d me on the letter, and I said, that was just a brilliant letter, it was perfectly worded. And she said, well, based on our conversation yesterday, I actually used AI to generate the first draft of the letter, and it came out perfectly.

So I think she’s probably more like most of my colleagues, where this is novel, something they’ve heard of, not necessarily employed. But I think once people use it for the purposes along the lines I’ve been saying already, they’re going to find it’s a really useful, potential timesaving tool.

Kevin Pho: So for those medical students listening to you, what are some ethical minefields that they should be aware of when using AI? Because AI, as you know, it evolves so quickly. We’re already so much further than we were when it was first introduced a couple years ago. So tell us about some ethical quandaries that medical students should be aware of when utilizing AI in their education.

Wes Boyd: I think a lot about ethical issues in AI, because in my new role as director of education at Harvard Medical School in bioethics, we’re thinking a lot about the role of AI generally, but also with respect to all things pertaining to medicine.

And my feeling right now is, the field is so new and its potential is so vast that we really can’t predict where things are going to be in 10 or 20 years with AI. So even trying to nail down what the current ethical issues are is really impossibly difficult in many ways.

I’m of several mindsets. One, I wonder if AI today is what the internet was 25, 30 years ago. Is AI today just a souped up version of what happened, say, 25, 30 years ago when the internet came on the scene for the majority of folks? And we had a lot of the same kinds of questions back then.

I also think about AI the way I do about genetic testing. And specifically, because I’ve had reason to think about this, what we can do today with genetic testing was unthinkable 50 years ago. So for example, if I was a sperm donor 50 years ago and had a child conceived through donor technologies, it would have been inconceivable for me 50 years ago that today I could be identified with the snap of a finger. Like, hands down, no ifs, ands, or buts, you’re the person who is my biological father.

If you push forward 50 years, my feeling is, if we’re getting our DNA tested through entities like 23andMe or other similar entities, we have no idea, truthfully, what the implications of that testing today on me might be 50 years from now. And I feel that’s a pretty good analogy with AI, like I can’t probably even remotely conceive of where things might be in 50 years. So what the ethical issues are today, as pressing as they might be, it’s a minefield.

But in terms of the potential risks, which is what you kind of came up with, or were intimating at, I think there are definitely risks of bias with anything that AI might generate as an answer. For example, the AI systems are only as good as the data that they’re trained with, and if they’re trained with certain data from, say, majority populations in various ways, they may end up overlooking issues within certain minority populations whose data were not fed into the AI system in the first place.

I think there’s always a risk of depersonalization of medicine. If I’m relying on computers and technology to arrive at a diagnosis, maybe I’m going to find it less important to actually sit with a patient and get to know them, do a hands-on exam, for example. That element of medicine is what makes medicine so amazingly appealing to me, sitting with people, getting to know them, hearing their stories. And if I’m relying on something like AI, I may inadvertently, or not so inadvertently, devalue that kind of connection.

And then I think also the risk of losing one’s ability to use their own clinical judgment. It’s so important to be able to feel a gestalt when you’re sitting with a patient, or thinking about a patient’s overall presentation. What is the gestalt, what are the assumptions I’m making, what are my likely diagnoses, prognosis, et cetera? This is true for relying on technology generally, but certainly with AI as well. You just lose the potential ability to have that overall sense of solid clinical judgment for any given patient.

Kevin Pho: Now, have you heard of cases, and it could be medical students or medical educators, where they use AI in a clearly unethical manner, that crossed a clear ethical boundary? Have you heard any cases like that?

Wes Boyd: I’ve not. It’s a great question, but I’ve not heard of any instances like that.

I have heard situations where AI has flat out made up references from the academic literature. I’ve heard instances where AI has generated references that just were entirely made up. And so whatever conclusions that AI may have generated, if they’re doing it on the basis of completely fabricated data, then we can’t trust those data.

So along those lines, knowing that there have been instances, at least that I’ve heard of, where AI has generated fabricated references just whole cloth, de novo, out of thin air, in the talk that I was preparing for the lab workers, for example, I actually asked AI to give me suggestions for improving lab workers’ mental health and well-being that were referenced in the scientific literature. And so I got a list of six or seven things, or suggestions, from AI, all of which had references. But knowing what I knew, I went back and triple checked that those references were actually legit and out there in the literature so that people could find them.

But in terms of just flat out fraud, I’ve not heard instances of that happening. And I know at least the purveyors of a lot of the open platforms in AI are doing everything they can to prevent that from happening.

Kevin Pho: Now, you mentioned earlier that some medical schools regulate the use of AI, some medical schools turn a blind eye to AI. What’s the happy medium there? Should medical schools address it head-on and create standards of education regarding AI? What are your thoughts on that?

Wes Boyd: The latter, absolutely, what you said. AI is here to stay. I mean, just like 25, 30 years ago, the internet’s here to stay. Medical schools need to acknowledge its existence, they need to embrace it, they need to know that their students are going to be using it, they need to know people like me on the faculty are also going to be using it, perhaps in different ways than students.

And then figure out how to ensure that students still learn to arrive at diagnoses on their own, and that students are still able to learn to talk with patients, sit with patients and hear their stories and empathize with patients, but also figure out ways proactively to use technology, including AI, to be as excellent as possible as physicians.

I mean, if I have a clinical judgment and I may be wavering one way or the other, which diagnosis is it, and I then go to technology to help me sort out in the end both what the diagnosis is and how I’m going to proceed as a physician treating this person, that’s a best case scenario.

So I think medical schools need to fully embrace AI, because its potential to help patients is dramatic. And you can do that, in my opinion, in a way that preserves people being able to render their own clinical diagnosis accurately, as well as possible.

Kevin Pho: We’re talking to Wes Boyd. He’s a psychiatrist. And today’s KevinMD article that he co-wrote with medical student Amelia Mercado is titled “The role of AI in medical education: Embrace it or fear it?” Wes, as always, we’ll end with your take-home messages to the KevinMD audience.

Wes Boyd: AI is here to stay, and we’ve got to do everything we can to both acknowledge its existence and also employ it for good. I think there are, as you alluded to earlier, Kevin, potential harms that might come from it. But I think by being as open eyed as possible about the fact that the future is already here, that’s the way we’re going to ensure that AI is used for good purposes, and to make future physicians even better than they would be otherwise.

Kevin Pho: Wes, thank you so much for sharing your perspective and insight, and thanks again for coming on the show.