Artificial intelligence is an increasingly valuable coin of the realm in medicine. It is already playing a role in disease diagnosis, drug development, surgery, maternal and child health, and even depression treatment, and some analysts believe that by 2026, AI could reduce annual health care costs in the United States by $150 billion. After a slow start, it is also gaining attention in lower-income countries as a way to improve service delivery and health outcomes. What other opportunities for using AI in medicine are likely to emerge? Does AI pose risks to safety and privacy, offer the promise of more patient-centered health care — or both?