AI in Med Ed: Treading Carefully While Giving Students the Right Tools

By Kelly Villella-Canton, Segment Leader & Director of Product Management, Medical Education & Practice, Wolters Kluwer Health
LinkedIn: Kelly Villella-Canton
X: @Wolters_Kluwer

Over the past year, generative AI – a type of AI technology that can produce high-quality content ranging from images to text to audio – has taken the globe by storm and is already being applied across industries including entertainment, law, finance and healthcare. For educators and students, generative AI has garnered their attention thanks to this technology’s uncanny ability to mimic human speech in writing and churn out answers to complex questions.

In medical education specifically, earlier this year, a study found that ChatGPT, a high-profile generative AI tool, was capable of passing the United States Medical Licensing Exam (USMLE), which consists of three separate tests. The researchers concluded that ChatGPT “demonstrated a high level of concordance and insights in its explanations,” which many believe opens the door for generative AI’s potential to support medical education.

However, while generative AI is exciting, this isn’t the first time AI has made headlines in the education space. For example, AI-enabled adaptive learning helps personalize educational experiences based on factors such as a student’s performance and preferences. As we approach and further refine the role of AI in medical education, here is what the industry will need to contend with.

Balancing innovation with risk

While cheating and plagiarism will always be concerns, many are still bullish about generative AI’s potential to advance education and offer new opportunities for learning. However, this is the first time students have had such easy access to a tool capable of generating new, original and potentially inaccurate content.

A major concern with generative AI, particularly in medical education, is that the content generated is not always pulled from evidence-based sources. Open models pull data from disparate sources and not necessarily from the peer-reviewed academic articles medical students would normally review. Furthermore, AI chatbots sometimes “hallucinate” by making up information entirely. In a field that relies on highly vetted, repeatable and reliable research, these potential inaccuracies could lead to patient harm.

To get ahead of this, medical schools and residency programs must consider smart ways to leverage generative AI tools that are carefully vetted and can incorporate the latest medical evidence. Doing so will ensure students have access to resources that could streamline their education, rather than damage their learning.

Giving students the tools, not the answers

It’s also important to remember that generative AI can’t take the place of the many years of learning needed to become a competent healthcare professional. Generative AI tools won’t be able to help a student in the exam room or when face-to-face with a patient. In order to thoughtfully integrate generative AI into medical education, the industry must consider appropriate use cases that don’t compromise the quality of a student’s learning.

As long as the content generated is reviewed beforehand by well-regarded experts, generative AI has the potential to make learning more efficient. With this technology available at students’ fingertips, it’s best to harness it into vetted tools that will support students without doing the work for them. For example, generative AI could be useful in helping tutor students by summarizing complicated topics or academic articles into a few lines of easily digestible text. Generative AI may also be useful in developing practice quizzes, which are typically time-consuming to create. However, these practice quizzes created by AI might still need to be vetted by experts, at least initially, until the AI is highly trained in specific topics.

In the long run, the benefits of generative AI in medical education may not come from opening up these tools broadly for unregulated student use, but rather from helping curriculum developers and educators channel this technology into scalable study aids.

Generative AI is here to stay

What will the future of medical education look like as the industry grapples with AI? For one, if content creators are able to leverage AI to develop these vetted tools, that could enhance medical education overall. Generative AI could allow for more opportunities for students to practice interactions with patients through virtual simulation, for example, allowing educators to give students more feedback at scale.

Generative AI was a leading topic in 2023. It’s clear this is no passing technology fad, and that generative AI will continue to impact education in both predictable and currently unforeseen ways. With this in mind, it’s important that students understand the limitations and benefits of these tools now so they can learn how to live with generative AI and apply it thoughtfully within their future practice.