The public tells us what it needs: through what it searches, posts, and now asks AI. My research listens and responds fast, published where medicine pays attention. The single overarching mission: work that saves lives.
Every year 350,000 Americans have a cardiac arrest outside a hospital; about 9% survive. We built ChatCPR, an open-source AI coach that out-instructed 911 dispatchers on real emergency calls.
We put real patient questions in front of ChatGPT and real physicians, then had licensed clinicians judge every answer blind. The chatbot's replies won 4 to 1.
Sportsbooks spread from 1 state to 38, and annual wagers grew from $4.9 billion to $121.1 billion. We tracked the fallout: Google searches for gambling addiction help are running at record highs.
Everyone complains about medical misinformation. Almost nobody tests a cure. So we studied Community Notes, the corrections volunteers attach to misleading posts, which have to cite their sources.
California passed two laws requiring social media platforms to default to censorship. This research served as key evidence that counterspeech already accomplishes what the laws proposed to mandate, so censorship would only displace a working solution. A federal judge struck down both laws in 2025.
PDFMy declaration in Kohls v. Bonta (E.D. Cal.) CN“Federal judge rebukes California laws targeting misinformation”