JAMA+ AI Conversations Maintaining Safety and Trust When Patients Engage Google: A Conversation With Dr Michael Howell
Sep 18, 2025
Dr. Michael Howell, Chief Clinical Officer at Google, dives into the intersection of AI and healthcare. He discusses the rapid advancements in AI, tackling the challenge of addressing 15% of novel search queries. Howell emphasizes the importance of balancing innovation with safety and outlines Google's commitment to rigorous testing and peer-reviewed guidelines. He also shares insights for early-career professionals in healthcare technology, advocating for a future where AI enhances patient empowerment and clinician support.
AI Snips
Chapters
Transcript
Episode notes
Search Queries Are Vast And Novel
- Human language queries occupy a vast, high-dimensional space, with 15% of daily searches never typed before.
- Models must handle queries that by definition cannot exist in their training data.
Compare Models To Real Clinical Baselines
- AI model outputs should be compared against realistic baselines like physician error rates.
- In some controlled tests, physicians had higher error rates than models on judged answers.
Patient Safety Roots Inform Google Work
- Michael Howell draws on his career in patient safety and quality to frame Google's health work.
- He views Google responsibility through the lens of preventing harm and improving outcomes.