How Future Doctors Are Using ChatGPT: Inside the AI-Powered Medical Classroom

Host 1: Okay, let's unpack this. Imagine a technology so powerful it can ace medical licensing exams,

Host 2: Mhm.

Host 1: um, seemingly overnight.

Host 2: Yeah.

Host 1: How does that reshape the very foundation of becoming a doctor?

Host 2: Mhm.

Host 1: Or even how you, you know, as a patient, experience healthcare?

Host 2: It's a huge question.

Host 1: Welcome to The Deep Dive.

Host 2: Yeah.

Host 1: Our mission here is, well, to cut through the noise, give you the most important insights, and leave you truly well informed.

Host 2: Mhm.

Host 1: Today, we're taking a deep dive into something vital and, uh, deeply personal:

Host 2: Mhm.

Host 1: the future of medical education and practice in the age of AI.

Host 2: And what truly stands out here is that we're not just, uh, talking about theoretical impacts, we're getting a firsthand account from people living this transformation right now.

Host 1: Right.

Host 2: Specifically, a recent med school grad who's also a venture capitalist, and a current medical student, both right there at the cutting edge.

Host 1: It's real experiences.

Host 2: Exactly. They've shared some really invaluable insights from, you know, their direct experiences, and we'll explore what it means for how doctors are trained, how they practice, and ultimately what it might mean for you as a patient.

Host 1: Okay, so our first guest, Morgan, he had this really fascinating journey into this whole space even before the, uh, public explosion of generative AI.

Host 2: Right, before ChatGPT was everywhere.

Host 1: Yeah. He moved from early computer science in high school to actually witnessing this monumental shift in hospitals. You know, the transition from old paper charts to complex digital systems.

Host 2: Oh, yeah, that was a huge change.

Host 1: And that firsthand experience, it really ignited his interest in how technology could, maybe, scale patient care far beyond the traditional one-to-one interaction.

Host 2: And then came what for many in medicine was a really pivotal moment. I mean, imagine dedicating years to grueling medical school,

Host 1: Right.

Host 2: all that work,

Host 1: Only to see a system like GPT-4 emerge and it's capable of acing the very licensing exams you're sweating over.

Host 2: Wow.

Host 1: Morgan described this as, uh, a humbling experience. Said it forced him to "interrogate what my role in medicine would be."

Host 2: That's quite a thing to grapple with.

Host 2: It really is. His personal "AI awakening," you could call it, it really underscores a crucial insight for future doctors. It's maybe less about memorizing tons of facts and more about defining their unique, irreplaceable human value in a world with these powerful AI tools.

Host 1: So, it's a fundamental shift.

Host 2: A profound redefinition of medical purpose, yeah. Not just humbling, but transformative.

Host 1: That introspection from Morgan is incredibly powerful, isn't it?

Host 2: Yeah.

Host 1: It pushed him to explore new paths, and ultimately, it led him towards genetics, a field he sees as, uh, uniquely poised for AI integration.

Host 2: Interesting choice.

Host 1: And our second guest, Daniel, he's a current medical student, and he kind of echoes this daily engagement with AI. Says he uses generative AI all the time.

Host 2: All the time, like, how?

Host 1: Well, everything from solving complex coding problems for his research, to simplifying really difficult medical concepts. You know, asking it to explain to me as if I was a 6-year-old.

Host 2: Huh, that's actually brilliant. Using it as a personal tutor.

Host 1: Pretty much.

Host 2: So Daniel's consistent daily use for coding, for decoding niche medical abbreviations, I mean, that just highlights how quickly these tools have become almost indispensable for the next generation of clinicians.

Host 1: Yeah, it really shows you the rapid pace of change, doesn't it?

Host 2: And the adaptability of these students. They just pick it up.

Host 1: So, okay, if AI is already this integral to how students learn and, well, function,

Host 2: Mhm.

Host 1: how is medical education formally adapting? Are the schools keeping up?

Host 2: Well, that's the interesting part. The answer, as both Morgan and Daniel pointed out, is that there's currently, uh, not a lot of integration of AI into the core medical curriculum.

Host 1: Really? Not much formal teaching on it.

Host 2: Not formally, no. Much of the innovation, Daniel notes, is actually student-led. They're figuring it out themselves.

Host 1: That brings up a critical point then about the, you know, the old ways of training versus where we're headed.

Host 2: Yeah.

Host 1: Morgan has a very nuanced view here. He believes it's crucial for future clinicians to understand the architecture of the medical note. You know, that foundational structure and reasoning behind how patient info is documented and communicated.

Host 2: Right, the fundamentals.

Host 1: Exactly. But, he argues, once that understanding is there, AI scribes should be introduced quickly, maybe in a matter of weeks.

Host 2: Wow, weeks. Those are the tools that listen in and draft notes automatically, right?

Host 1: Yeah. He jokes it's like learning cursive, you learn it, but then you move on to more efficient tools, like typing.

Host 2: Huh, okay. I see the analogy.

Host 1: Put that's a fascinating comparison. But Daniel, you know, from his perspective as a current student, he shares a different caution.

Host 2: Oh.

Host 1: While he definitely sees the value in tools like ambient note taking, he stresses that there's a lot of learning when you write the note. There's something about that process itself.

Host 2: Mhm, I can see that.

Host 1: He fears that relying too heavily, too early maybe, on AI, might hinder the ability to critically think, a skill that, you know, really requires practice to develop properly.

Host 2: That tension Daniel highlights is absolutely crucial. If thinking and synthesis actually happen during the note-taking process itself, and I know an expert from Epic, the big electronic health records company, noted something similar,

Host 1: Right.

Host 2: then automating that entirely for a student could, indeed, be counterproductive. It's a real dilemma, isn't it? Balancing that efficiency gain with the need for deep learning.

Host 1: It really is. But despite that formal curriculum lag, these students are, as you said, undeniably voracious adopters.

Host 2: They're using it anyway.

Host 1: Totally. Morgan shared a funny real-world example.

Host 2: Mhm.

Host 1: He uses AI tools like OpenEvidence or GPT for "defense against pimping."

Host 2: Against pimping? You mean when attendings grill them on rounds?

Host 1: Exactly. That rapid-fire quizzing by a senior doctor, AI gives him instant answers.

Host 2: Wow, okay, resourceful.

Host 1: And Daniel confirmed that many students are definitely consulting AI, especially for, like, a second set of eyes on complex cases,

Host 2: Mhm.

Host 1: or maybe to get the rationale behind why a doctor might order certain labs.

Host 2: So they're using it as a learning tool, a check.

Host 2: That makes sense.

Host 2: But building on that point about individual adoption, what happens when the institutions, like the hospitals, try to push back?

Host 1: Good question. Morgan mentioned a Rhode Island hospital that tried to block AI tools.

Host 2: Okay, so they banned it.

Host 1: Tried to. He found a way around it, just used his phone.

Host 1: And his take is pretty insightful.

Host 2: Mhm.

Host 1: He suggests that transparency encourages good behaviors much more than outright bans.

Host 2: Ah, so banning just drives it underground.

Host 1: Exactly. And that could lead to less responsible use, potentially. Better to have it out in the open and guide its use.

Host 2: Makes sense.

Host 2: Okay, so looking ahead then, how does AI reshape the very structure of medicine itself, bigger picture?

Host 1: Well, Morgan observes that the field of AI research is kind of moving beyond just benchmarking AI against human performance. That was the old game.

Host 2: Right, like, can it pass the test?

Host 1: Yeah. Now, he notes that AI "collapses medical specialties onto themselves."

Host 2: Collapses them. What does that mean exactly?

Host 1: It means AI can offer a broader, more triangulated view of a patient's condition. It sees connections that maybe a single specialist, focused on their area, might miss.

Host 2: Ah, okay. So it integrates information across specialties. That's powerful.

Host 1: Very.

Host 2: Zooming out a bit then, this suggests that medical specialties themselves might need to evolve, right?

Host 2: Just like new tech has reshaped entire fields before.

Host 1: Seems likely.

Host 2: And Morgan also envisions board certification, you know, how doctors get licensed in their specialty, moving away from just rote memorization.

Host 1: Yeah, less about recall.

Host 2: Less about recall, more towards adaptive assessment. Testing a doctor's ability to effectively use the advanced tools at their disposal, competence with the technology, not just knowledge without it.

Host 1: That's a big shift in how we measure expertise.

Host 2: Huge.

Host 1: And Daniel, he's already experienced one of the biggest shifts directly with patients.

Host 2: Mhm.

Host 1: He mentioned people coming into the emergency department armed with AI-generated lists of possible diagnoses they've looked up online.

Host 2: Oh, I bet. Dr. Google on steroids.

Host 1: Pretty much. But his approach is key: he engages in shared decision-making. He acknowledges the list, looks at it with them,

Host 2: Okay, respectful.

Host 1: but then also provides that crucial clinical context and human judgment that the AI, you know, obviously might miss.

Host 2: That really makes you think, doesn't it?

Host 2: In a world where patients are increasingly informed, or maybe misinformed, by AI, how do you build that essential element of trust? Especially when, say, a second-year medical student is on the care team?

Host 1: Yeah, that's tough.

Host 2: Daniel emphasizes that transparency and just candid conversations are vital. It's about being up-front, acknowledging the patient's research, and then clearly explaining the clinician's unique value-add, the human element.

Host 1: Right. So, okay, the big question then: with all this incredible capability, will AI eventually replace doctors entirely?

Host 2: The million-dollar question.

Host 1: Daniel's take is insightful here. He says "replace" is a strong term.

Host 1: He doesn't see doctors becoming completely obsolete.

Host 2: Okay, so not obsolete, but...

Host 1: But rather a shift in what it means to be a doctor. The role changes.

Host 2: And what truly stands out there from Daniel's description is that the future physician might pivot. Less focus perhaps on being solely diagnostic,

Host 1: Because AI can help with that.

Host 2: Exactly. And more focus on being highly skilled at, quote, "understanding the limitations of these models and knowing when to intervene."

Host 1: Ah, the judgment calls. Knowing when the AI is wrong or missing something.

Host 2: Precisely. Even in surgery, you know, you have these advanced robotic systems like the da Vinci robots, they're great for certain precise tasks,

Host 1: Amazing technology.

Host 2: but a human surgeon is still absolutely essential to react to the unexpected, to make those critical judgments in the moment. The tools enhance, they don't fully replace that human expertise.

Host 1: Right, that makes sense.

Host 1: This leads us to a pretty powerful vision for the future then. Morgan, with his unique physician-plus background, medicine and venture capital...

Host 2: Yeah, that dual perspective.

Host 1: He predicts that in just 2 years, we'll see even greater adoption of AI, both for back office tasks, like admin stuff, and for patient-facing applications.

Host 2: 2 years, okay. That's fast.

Host 1: Yeah.

Host 1: And in 5 years, he expects regulators will have figured out payment mechanisms for AI, recognizing that, you know, not all valuable AI applications directly generate revenue through the old billing codes.

Host 2: That's a huge hurdle, the payment piece.

Host 1: Yeah.

Host 2: If you can't bill for it...

Host 1: Right, how do you sustain it?

Host 2: So, the most ambitious vision, the one Morgan hopes to see in maybe 10 years,

Host 1: Well...

Host 2: what is it?

Host 1: It's the dissolving of the barrier that exists between care delivery and biomedical discovery.

Host 2: Wow, okay. Break that down.

Host 1: It basically means learning from every single patient's experience, automatically, to improve care for the next patient. A true learning health system made possible by AI, constantly adapting and improving.

Host 2: So every patient encounter feeds back into the knowledge base, refining treatments.

Host 1: Instantly, potentially. And Daniel, thinking about his own future, he clearly hopes to blend that direct patient care with these higher-level tech projects.

Host 2: Yeah, it shows a generation that's really ready to build that future, doesn't it? They see tech as part of the job.

Host 1: Absolutely. What an incredible deep dive into the, uh, the front lines of AI and medicine.

Host 2: Really fascinating stuff.

Host 1: From Morgan's journey, you know, the VC and physician-in-training, to Daniel's daily experiences navigating med school with AI,

Host 2: Yeah.

Host 1: it's just clear that the next generation of clinicians is already deeply engaged in shaping this revolution.

Host 2: Yeah, and a key insight here, I think, is that while the accreditation bodies and the formal curricula are, let's say, slowly adapting...

Host 1: Need to catch up.

Host 2: Right, much of the innovation in how AI is actually integrated into medical practice and education is emerging directly from the students themselves.

Host 1: They're the pioneers.

Host 2: They really are. They're the ones hands on, experimenting, finding what truly works out there in the real world.

Host 1: So, okay, what does this all mean for you listening?

Host 2: Yeah, the takeaway.

Host 1: It means the future of your healthcare could be fundamentally different, potentially more personalized, uh, maybe more efficient, perhaps even more precise.

Host 2: Hopefully, all of the above.

Host 1: Guided by doctors who are not just experts in medicine, but also masters of these powerful new tools.

Host 2: Which leaves us with a pretty provocative thought for you to consider, doesn't it?

Host 1: What's that?

Host 2: In a world where AI can offer a nearly perfect diagnostic answer, or maybe a comprehensive treatment plan, what then becomes the true measure of a great doctor? What's left when the machines handle the data?