Move Over LLMs! Yann LeCun & Alex LeBrun Debut AMI Labs’ World Models for Healthcare

Are large language models enough for medicine? In this special episode of How I Doctor, Turing Award winner Yann LeCun and Nabla founder Alex LeBrun join Dr. Graham Walker to introduce AMI Labs and make the case for a new AI paradigm in healthcare: world models. While today’s LLMs power clinical documentation, chatbots, and decision support tools, Yann argues they are fundamentally designed to predict the next word — not simulate physiology or reason through consequences. In medicine, where reliability matters more than creativity, that distinction is critical. Together, they explore why stochastic AI systems may hit a wall in healthcare, how physicians mentally simulate patient outcomes before acting, and why training AI purely on notes reduces biology to language. This conversation isn’t about replacing doctors. It’s about building AI that can reason, plan, and amplify clinical judgment. If you’re a physician thinking about the future of AI in medicine, this episode is essential. ⏱ Topics Covered 00:00 – LLMs Predict Words. World Models Predict Patients 05:10 – What Is a World Model? 10:40 – The ER Mental Model Analogy 18:20 – Why 80% Accuracy Fails in Healthcare 24:35 – Language Is a Projection of Biology 31:50 – From Documentation to Real Reasoning 38:15 – The Future of AMI Labs and World Models 🔗 Episode Links Article + Transcript: offcall.com/learn/podcast/ai-world-models-medicine-yann-lecun-alex-lebrun Nabla: https://www.nabla.com/ Offcall: https://www.offcall.com Follow Dr. Graham Walker:   / grahamwalker   Follow Yann LeCun:   / yann-lecun   Follow Alex LeBrun:   / alexandrelebrun   #healthcareai #LLMs #WorldModels #medicalai #clinicalai #physicianlife #futureofmedicine #digitalhealth