Should You Trust ChatGPT With Your Health?
a16z partner Jay Rugani talks with Karan Singhal, who leads health AI research at OpenAI, about how ChatGPT handles health questions, why trust matters, and how AI could support patients and clinicians without replacing doctors.
The gist
-
Singhal says more than 300 million people use ChatGPT weekly for health questions, mostly free users outside the US, often between short doctor visits.
-
Rather than trusting a few doctors, OpenAI pools input from hundreds of physicians worldwide and trains models to urge care even when users want reassurance.
-
Singhal notes new clinical guidelines take 5 to 17 years to reach everyday practice, and says AI can bring current research into a 15-minute visit.
-
He says GPT-6 Luna beat the previous best model on HealthBench Professional at 34 times lower cost, and AdventHealth reported roughly 80% lower admin costs.
-
Singhal says AI won't replace doctors' empathy or exams, and expects its biggest impact in 5 to 10 years from new treatments and faster trials.
Also mentioned
4:18
Apple Notes
· "I saw a dermatologist last week about something. I wrote down all my questions beforehand in Apple Notes..."
5:06
The Cognitive Revolution
· "He hosts the Cognitive Revolution podcast and told his story in that podcast... I don't know if I'm allowed..."
11:12
Penda Health
· "A series of experiments that we ran with Penda Health where we studied this idea of a clinician co-pilot..."
15:26
HealthBench
· "Back even for the original HealthBench evaluation that we released back in May 2025, there was actually a..."
25:15
ChatGPT for Healthcare
· "We need to make sure we're able to offer HIPAA compliance. We need to make sure that we're able to offer..."
◎
Related episodes
Episodes covering similar topics
More from a16z
Agentic Commerce: Why Your Wallet Is About to Change
Oct 15, 2026
Recursive Self-Improvement and the Future of AI
Oct 12, 2026