The diagnostic excellence movement has largely been based in retrospective error adjudication, with the hope that by identifying trends in medical errors we can make specific systems changes to prevent them in the future. Generative AI models, especially with recent advances in "reasoning" models and agentic frameworks, offer a new paradigm -- identifying errors before they happen. This talk will describe a fundamental tension between two different care models — dyadic and triadic care — and detail technological innovations like scalable oversight and benchmarking that could theoretically upend diagnostic safety as we know it. The talk will also soberly evaluate the current evidence, with a focus on clinical trials.
Speaker: Adam Rodman, Assistant Professor, Harvard Medical School Director of AI Programs
Register
Event details
