Christopher Migliore
Assistant Professor of Medicine, CUIMC
For medical students, class sizes range from 2, when I’m on the wards, to 4, when I’m precepting, to classrooms of up to 70, when lecturing on a topic for the graduating seniors as they prepare for residency. For residents, it can range from groups of 1 if I am on a particular service, up to 30-40 if speaking at noon-report or learning lab. The learning goals are varied across these situations, but a consistent thread is translating clinical knowledge into clinical practice – how do you take book knowledge in medicine and translate that into excellent care for the patient.
Approach to teaching and learning in the age of AI
It’s important to understand that AI is rapidly becoming widespread in how learners approach their material. They are using it to simplify (“Explain the Krebs’ cycle like I’m a 1st year biology student”), and also using it to look up material more often (genuinely curious if Wikipedia has a future). While the material is usually correct, they sometimes seem blinded to the fact that there may be hallucinations. I remind them that 1000 correct answers in a row does not guarantee it won’t hallucinate on the next query. When discussing AI with my students, I ask them how they plan to use it, specifically. I also ask them to be honest when it comes to any assignments. I for one am not disappointed or upset when they utilize it to help provide the ‘skeleton’, which they then flesh out on their own. I’ve found that when this is addressed upfront, and the expectation is that the bulk of the work is done by the learner, they feel comfortable using AI more. It is an amazing assist, and my expectation is that don’t use it beyond that
Course elements that evolved with the integration of AI
Syllabi and structure of my curriculum goals has been made easier by ChatGPT (there is a plug-in that constructs diagrams, which makes visualization of learning goals across time, ensuring curricular threads don’t fall off, etc, much easier to visualize). For example, if I have 5 large learning goals for the block, I can have ChatGPT construct a mindmap which can help learners focus on the big goals, create their own goals underneath the main ones, and then include an area for their personal expectations of the block. Personalizing goals for learners was hugely time-consuming before; AI has simplified this and my learners seem more invested, since part of the map is their creation. As far as evaluations, while they always require revision, asking chat-gpt to generate 50 question stems simplifies the assessment process. I think AI can best be summed up when it comes to assessment in ‘well-begun is half done’. Before ChatGPT, question writing was much more daunting. Having so much raw material spit out so quickly not only speeds up the process but decreases my energy of activation needed to work on it at all.
Teaching students to leverage AI and develop their AI literacy
One of the main tools I encourage medical students and residents to become familiar with is openevidence.com. This is an AI-powered medical aggregator, where questions can be entered in conversational English and answers are given in global English with references provided below. The details of the ‘guts’ seem misty, but as it is more an aggregator and less a deep-learning machine, hallucinations are less of an issue. Apart from this, I encourage my residents with a quote I heard at a talk by Atul Gawande, “As physicians, your job will not be replaced by AI. Your job will be replaced by someone adept at AI.” Occasionally I encounter a learner who is less familiar with AI. If they’re interested in learning more, I have them start small, with openevidence.com. This is a Mayo Clinic creation which provides cleanly written summaries of the current medical literature after the user types in a question. It exposes them to material that is easily checkable, helps them write their assessments and plans, and (hopefully) decreases any unease they feel with AI in general.
Lessons learned from teaching and learning with AI
From teaching and learning with AI, I’ve learned that it can foster personalized, at the level you’re at, education. AI, when leveraged properly, can offer real-time feedback, allowing learners to grasp complex concepts more swiftly. From my perspective, and most importantly, using AI encourages adaptation to technological advancements. I want my learners to feel comfortable with the arguably black-swan level of disruption AI will cause.
Advice for colleagues on leveraging AI for teaching and learning
Dive in and test its limits. If there’s an area of your course or classroom experience you’re struggling with, see if you can leverage it to your advantage. It’s not going away.