Will AI Replace the Medical Curriculum? Inside the World Bank's Latest Education Data

Sam: Imagine finding out that 81.1 percent of medical students have used an AI-powered tool to learn how to be doctors, but almost half of them have never been formally taught how to use it. That is the exact reality we are looking at today.

Maya: It is a massive wake-up call for health systems and tech companies. Today we are unpacking a May 2026 implementation study published by The World Bank titled ARTIFICIAL INTELLIGENCE INTEGRATION IN MEDICAL EDUCATION: A Landscape with HOT Dimensions in Viet Nam.

Sam: Just a quick reminder before we get into the data, our voices are AI-generated, and this podcast is a summary of a publicly available document. Also, nothing we discuss today is medical advice.

Maya: Exactly. And this document is fascinating because it gives us a massive, empirical look at what is actually happening on the ground. The authors, Sang Minh Lê and Giang Bảo Kim, looked at a sample of 2,213 medical students and 600 teachers across six selected medical schools.

Sam: And the reason they looked at Viet Nam specifically is really interesting. The country's data economy is exploding. The report notes the digital economy accounts for about 14 percent of GDP and is growing by over 20 percent annually. They have incredible infrastructure growth, like nationwide fiber optic coverage exceeding 82% of households.

Maya: Right, and in the education sector, the pandemic really pushed things forward. As of 2023, more than 50 percent of higher education institutions delivered programs partly or fully online, and 75 percent of Vietnamese students accessed online learning. So the digital tracks are laid.

Sam: But when we pivot to medical education specifically, the World Bank found a landscape that is full of contradictions. Let us start with the sheer scale of adoption. The numbers are staggering. Among the selected medical schools, 81.1 percent of students and 69.9 percent of teachers have used an AI-powered tool.

Maya: The frequency is high too. The authors found that 70.4 percent of students and 58.1 percent of teachers frequently adopt AI in their learning and professional activities. But here is where you have to look closely at the data. When you ask about daily use, the numbers drop sharply. Only 9.8 percent of students and 14.9 percent of teachers use these applications on a daily basis.

Sam: That drop-off is telling. It suggests people are trying it out, finding it useful for specific tasks, but it is not completely running their day-to-day lives yet. And what exactly are they using? The report is very clear that they are not mostly using highly integrated, enterprise-level medical software.

Maya: Exactly. AI-powered learning assistants, things like ChatGPT or Watson Assistant, are by far the most widely adopted tools. We are talking about 77.6 percent of male students and 76.3 percent of female students using them. For teachers, it is 67.7 percent of males and 61.1 percent of females. The authors emphasize that AI is primarily utilized as an individual productivity tool rather than an embedded institutional resource.

Sam: So it is a bottom-up revolution. The universities are not handing this technology down from on high; the students and teachers are just pulling it off the internet to make their lives easier. What are they actually asking the AI to do?

Maya: For students, it is heavily skewed toward knowledge acquisition. Knowledge searching and synthesis is the top function, reported by 75.3 percent of male students and 64.9 percent of female students. They are using these tools to summarize huge lecture slides, synthesize complex topics, and generate practice multiple-choice questions.

Sam: The report even includes some direct quotes from students that are incredibly relatable. One student mentioned copying their entire slide outline, pasting it into a chatbot, and asking for a summary. They said they only go back to the original slides if they find something they do not understand in the summary.

Maya: Which is efficient, but also highlights a major risk if the AI hallucinates, which we will get to later. Interestingly, they also use it to pull information for clinical decision-making. The report shows 29.0 percent of male students and 23.4 percent of female students are doing this. It shows AI is starting to link theoretical knowledge with practical clinical contexts.

Sam: And the teachers? Are they using it to grade papers or build courses?

Maya: Mostly for prep. Nearly half of male teachers, 47.5 percent to be exact, and 41.9 percent of female teachers use AI for the design and development of lesson plans. Compiling books and learning materials is also huge, utilized by 46.0 percent of male teachers and 35.6 percent of female teachers.

Sam: But creating actual curricula or evaluation tools is lower, right?

Maya: Much lower. Using AI for the design and development of curricula is only reported by 7.2 percent of male teachers and 14.1 percent of female teachers. And using it for clinical skills training or education management remains relatively limited, with adoption rates around or below 10 percent for many of those specialized functions.

Sam: This perfectly sets up the core framework of the document. The authors break the barriers to adoption down into three dimensions: Human, Organization, and Technology, or what they call the HOT dimensions. Let us dive into the Human factors first. How do people actually feel about all this AI in the classroom?

Maya: Generally, the attitudes are very supportive. The report notes that over 90% of students and 84% of teachers believe that AI will become an essential tool for teaching and research within the next 5 to 10 years. They see massive benefits. For example, 73.5 percent of male students and 69.1 percent of female students report perceived skill improvements in knowledge searching and synthesis.

Sam: And teachers see it as a way to boost their own capacities too. The data shows 52.4 percent of female teachers and 43.4 percent of male teachers perceive improvements in their technology use thanks to AI. So the enthusiasm is absolutely there.

Maya: Enthusiasm, yes, but also a healthy dose of fear. The report found that 43.5 percent of students and 28.9 percent of teachers suppose the adoption of AI has both positive and negative impacts to the quality of medical education. Stakeholders described a cautious optimism.

Sam: What are they most worried about?

Maya: Over-reliance is the big one. There is a deep concern about students becoming dependent on AI, particularly regarding clinical reasoning skills. If the computer does all the connecting of the dots, the student might lose the ability to think critically. Students themselves are also worried about irresponsible use, feeling that AI increases the risk of plagiarism and undermines fair evaluation during online assessments.

Sam: One student in the focus groups complained about classmates just pasting prompts into a chatbot and copying the output directly for group work, without adding their own ideas. That has to be maddening. It also raises questions about who exactly is using this the most. Are there demographic splits?

Maya: Yes, significant ones. Gender and age are important determinants. The data shows male students and male teachers tend to use AI more frequently. The percentages of male students and male teachers using AI daily are higher than those of female students and female teachers with statistical significance. Also, teachers under the age of 40 and with fewer than 10 years of work experience demonstrate higher levels of AI adoption.

Sam: That generational divide makes total sense, but it also creates a management problem. If the senior faculty are hesitant to adopt AI, they are going to struggle to create policies or curricula that effectively guide the younger, highly adopting students. Which brings us perfectly to the second dimension: Organizational factors.

Maya: This is where the alarm bells really start ringing. The adoption is high, but it is entirely informal. A staggering 46.8 percent of students and 47.2 percent of teachers have never received formal training in AI. They are just figuring it out on their own.

Sam: Where are they even learning to use it if the universities are not teaching them?

Maya: Quick note before we carry on. This spot is open for a sponsor. If your company builds or sells AI for healthcare and wants to reach the clinicians, health-system leaders and industry teams who listen to this show, the link to our sponsorship page is in the show notes.

Sam: And now, back to the document.

Maya: The internet and social media. The report says 67 percent of students learned about AI through the Internet, and 57 percent through social media platforms. For teachers, it is similar: 45.8 percent through the Internet, 32.9 percent on social media, and 31.6 percent from colleagues.

Sam: That is wild. We are talking about training future physicians, and the primary mechanism for learning how to use an advanced clinical and educational tool is social media. That lack of structured training has to cause problems on the ground.

Maya: It absolutely does. The authors note that 61.7 percent of students and 31.6 percent of teachers reported experiencing difficulties when using AI. The most prominent challenges stem from this lack of formal training, limitations in technological skills, and, crucially, insufficient financial resources.

Sam: Let us talk about the financial piece. Good AI is not always free, especially the models that are less prone to hallucination or that can handle large medical documents.

Maya: Right, and medical schools generally do not provide financial support to cover paid-version AI subscriptions. So, 71.1% of students use free tools, while only 28.9% reported using at least one paid AI version. Teachers are a bit more willing or able to pay, with 43.9% using at least one paid version, compared with 56.1% using only free versions.

Sam: That creates a massive equity issue. If a wealthier student can afford a premium model that summarizes text perfectly and never hallucinates, and a lower-income student relies on a free version that makes up fake citations, the playing field is entirely uneven.

Maya: Exactly. The report specifically notes that female students and teachers reported greater difficulty in accessing adequate financial resources and equipment compared to their male counterparts, with statistically significant differences. The authors highlight this as a gender disparity that needs addressing.

Sam: So if I am a tech company or a vendor selling into this space, I need to realize that high subscription costs are pushing the vast majority of the academic market onto free, potentially lower-quality tiers. There is a huge opportunity there for discounted academic licensing.

Maya: The students think so too. One student quoted in the report specifically asked for developers and companies to offer discounted packages for academic users. But cost is not the only organizational barrier. There is also a massive cloud of ethical and legal uncertainty. Ethical and legal risks are cited by nearly half of teacher respondents, specifically 46.8 percent of males and 48.3 percent of females.

Sam: I would be terrified as a faculty member too. If a student feeds real, de-identified patient data from a clinical rotation into a public chatbot to help write a case report, who owns that data? Is it a privacy breach? The universities have not set the rules of the road.

Maya: You hit the nail on the head. A medical school leader in the seminar discussion pointed out the risk of accessing patients' medical records to extract information without consent. They noted there is no comprehensive legal framework for AI, making its application carry a relatively high level of risk. This regulatory uncertainty discourages medical schools from fully embracing AI.

Sam: Which brings us to the final dimension: Technological factors. We already mentioned that Viet Nam has a booming digital economy, but having fast internet at home does not mean the hospital systems are ready to plug into advanced AI, does it?

Maya: Not at all. The report identifies digital data underdevelopment as the most critical barrier to AI integration in medical education. Medical institutions still rely heavily on traditional formats like paper-based examinations. The costs and time required to convert hard-copy data into digital formats are major obstacles.

Sam: And that extends to the clinical side as well, right? The actual hospital data.

Maya: Yes. The Ministry of Health had a goal for nationwide implementation of electronic medical records by September 2025. But according to the report, only 881 out of 1,645 hospitals had adopted EMRs as of the deadline. That is just 53.6%. You cannot run advanced, AI-driven learning analytics or clinical simulations if half the hospitals are still using paper records.

Sam: So we have this weird paradox. Students are sitting in hospitals with paper records, pulling out their smartphones over a 5G network to ask a generative AI chatbot about a disease. It is a clash of two totally different technological eras.

Maya: It is, and that clash creates real dangers regarding AI accuracy and bias. Students frequently report that the information provided by AI is inaccurate, not evidence-based, or overly general. One student mentioned that the AI has not yet incorporated official Vietnamese-language sources, like regulations and decisions issued by the Ministry of Health.

Sam: That is a huge takeaway for companies building AI tools for global markets. If your model only trains on English-language medical literature and ignores local clinical guidelines or Ministry of Health protocols, it is practically useless, or even dangerous, for a medical student trying to learn the standard of care in their specific country.

Maya: Precisely. The authors note that concerns about inaccuracy and hallucinated references seriously undermine the credibility of these tools in a field that demands high precision.

Sam: So, how does the World Bank suggest we fix all of this? What are their final recommendations for getting out of this fragmented, informal phase?

Maya: They lay out a few key strategic actions. First, systematic capacity-building programs. Medical schools absolutely must introduce structured AI literacy training that covers ethical considerations and practical applications, so students aren't just learning from social media. This is especially important for senior faculty to bridge that generational divide.

Sam: Makes sense. You cannot regulate what you do not understand.

Maya: Second, AI needs to be systematically embedded into medical education curricula. It should be integrated into competency-based education models. They need practical modules for AI-assisted clinical decision-making and research.

Sam: Right, move it from being a secret shortcut to an openly evaluated competency.

Maya: Exactly. Third is institutional governance. Ministries and universities need to establish clear guidelines on ethical use, data privacy, and academic integrity to remove that regulatory fear. And finally, they must strengthen the digital data ecosystem by digitizing educational data and pushing for interoperable electronic medical records.

Sam: Plus, fixing the financial aspect. They recommend universities allocate budgets for licensed tools or partner with tech companies for academic pricing to ensure equitable access.

Maya: If I had to pick one memorable takeaway for health-system leaders and tech companies from this document, it is this: AI is already in your medical schools and hospitals. The students have brought it in through the back door as a personal productivity tool. The challenge now is bringing it out into the light, funding it, and building a formal curriculum around it before the bad habits and data risks become permanent.

Sam: That is the perfect summary. The adoption phase is over; the integration phase is what matters now. To recap today's episode: A World Bank study of 2,213 medical students and 600 teachers in Viet Nam shows massive, but highly informal, AI adoption. 70.4 percent of students frequently use AI tools, mostly free chatbots, but nearly half have no formal training. Moving forward requires urgent investments in digital infrastructure, clear ethical guidelines, and formalized AI curricula.

Maya: As always, you can find a link to the full World Bank document in our show notes. And remember, this podcast is for informational purposes only and is not medical advice.