Purpose
AI in medical education raises fears regarding teaching challenging concepts effectively and the reduction of critical thinking. The gap between AI's potential and implementation highlights a need for a framework to guide how and when to apply AI in curricula. We adopted the PICRAT model to determine when and how to use AI to create activities and explain challenging concepts/ misconceptions. 

Methods
We evaluated the application of AI (ChatGPT 4o and Claude 3.5 Sonnet) in medical education using the PICRAT model, which assesses technology's educational impact on two axes: student engagement (Passive, Interactive, Creative) and teaching practice influence (Replacement, Amplification, Transformation). This work represents a student-educator perspective and collaboration that discussed a framework for applying AI to explain complex concepts like gradients, flow, and velocity and Darrow Yannet diagrams or fluid-electrolyte balance visualizations.

Results
Two key findings emerged regarding AI limitations and effective integration. While effective at concept visualization, AI requires human input for comprehensive explanations, struggling with nuances like steady-state versus dynamic changes. Second, applying the PICRAT framework ensures that AI use augments critical thinking. Applications are not replacements (e.g., interactive figure for a paper figure) and are student-centric learning.

Conclusion
AI should complement, not replace, human teaching and learning. Our modified PICRAT-guided integration can guide the selection of when and when not to use AI to promote understanding and minimize overreliance. The goal is that technology amplifies or transforms education rather than simply replacing existing practices.