NEJM AI Grand Rounds What Values are in AI? A Conversation with Dr. Zak Kohane
21 snips
Dec 17, 2025 Dr. Zak Kohane, Chair of Biomedical Informatics at Harvard Medical School, shares his personal journey with AI in medicine. He discusses building AI tools for clinical use and highlights how AI is rapidly transforming reimbursement practices, especially in revenue-cycle optimization. Zak emphasizes the importance of understanding value judgments in AI models, which can differ from human clinicians. He calls for new oversight to address biases and align AI with patient interests, stressing the need for transparency and ethical deployment.
AI Snips
Chapters
Transcript
Episode notes
Building A Personal EHR With AI
- Zak Kohane built a personal EHR in about six hours using AI-augmented coding tools to integrate scattered medical data and scans.
- He uses an AI-trained model on his Gmail to draft replies and to suggest content, speeding personal workflows.
Money Drives Faster AI Adoption Than Clinical Use
- Despite early expectations, radiology, dermatology, and ophthalmology haven't broadly adopted AI for interpretation at scale yet in 2025.
- By contrast, revenue-cycle AI and upcoding tools have expanded rapidly because financial incentives drive adoption.
Ambient AI Bought For Clinician Satisfaction
- Hospitals are investing heavily in ambient documentation AI, not because it increases measurable productivity, but because it improves clinician satisfaction.
- This suggests adoption can be driven by clinician well-being even without direct reimbursement incentives.