Podcast: AI CPR coach near-aces checklist on 911 calls, human dispatchers score lower
mdredze · x · 2026-11-21
On The Doctor's Lounge, UC San Diego epidemiologist John Ayers discusses evaluating healthcare AI by real health outcomes rather than test scores. His study of an AI system coaching bystanders through CPR during 911 calls found it met nearly all CPR checklist criteria while human dispatchers scored lower. He covers how the agent was built and tested, why real-world evaluation matters, and its potential in emergency response.
More from Apps
- ChatGPT Space rolls out to Plus users with collaboration, mobile editing — gdb · 2026-11-21
- Google AI Studio revamps signed-out landing page to showcase models and apps — ammaar · 2026-11-21
- Cargo's GitHub integration brings PR-style code reviews to sales and marketing workflows — cneuralnetwork · 2026-11-21
- Mitchell Hashimoto's Superlogical names its AI product Rex, opens Mac beta invites — aronchick · 2026-11-21
- Google's personal agent limited to one Gmail account, criticized as an own-goal — gaganghotra_ · 2026-11-21
- User reports AI device "dot" caught credit card fraud and gave full handling instructions — ilge · 2026-11-21