Summary
In this a16z podcast, healthcare and biotech partner Jay Rughani speaks with the head of health AI research at OpenAI about how people use ChatGPT for medical questions and how the company is trying to make those answers trustworthy.
OpenAI reports that more than 300 million people each week use ChatGPT for health and wellness questions. The conversation explores why: long waits for specialist appointments, short visits that leave patients with unanswered questions, and the gaps between appointments where AI can help people prepare, follow up and navigate their care. The guest describes OpenAI’s goal of raising both the floor and the ceiling of human health, from free access to reliable information to accelerating medical science.
The discussion turns to trust and safety. Topics include training models to recognize their own uncertainty, how HealthBench evaluates behavior such as hedging, and how licensed clinicians and mental health experts help define model behavior in sensitive scenarios. They also cover how to handle contested topics like unapproved peptides and vaccines by relying on expert-defined standards. The episode closes on designing products for patients, individual clinicians and healthcare enterprises, including the model, harness and deployment choices involved.
📺 Source: a16z · Published October 20, 2026
🏷️ Format: Podcast
