Medical AI evaluation rigor is crucial for safe deployment. Aneesh Gupta explains why current benchmarks fall short and how ...