# Can AI actually save lives? > A physician explains how AI is helping in the doctor's office and the lab. - Title: Can AI actually save lives? · Vox | Shortify - Summary: A physician explains how AI is helping in the doctor's office and the lab. AI is already saving lives in healthcare through administrative tasks, patient… - Keywords: artificial intelligence, healthcare, drug discovery, medical technology, clinical decision support, dhruv khullar, weill cornell medicine, ai scribes, patient navigation, lead optimization, clinical trials, ai regulation - Source: Vox — https://vox.com/podcasts/503589/ai-healthcare-patient-experience-drug-discovery - Author: Peter Balonon-Rosen - Published: 2026-09-23T19:50:00+00:00 - Read time: 3 min - Topics: artificial intelligence, healthcare, drug discovery, medical technology, clinical decision support, dhruv khullar, weill cornell medicine, ai scribes, patient navigation, lead optimization, clinical trials, ai regulation ## Americans trust AI in medicine While 70 percent of Americans worry about AI overall, they're optimistic about its use in scientific research and medicine. This is the one area where AI has broad public support. ## AI is already saving lives Dr. Dhruv Khullar, a physician at Weill Cornell Medicine, says AI has started saving lives and is being adopted faster in healthcare than any previous technology. The healthcare system's problems create appetite for change. ## AI scribes transform doctor visits AI is rapidly taking over administrative tasks like note-taking during patient visits. Doctors can now look patients in the eye instead of typing, improving the quality of interaction. ## AI helps patients navigate care AI tools help patients with serious illnesses like cancer or heart failure navigate complex healthcare systems. They answer questions about appointments, medications, and next steps seamlessly. ## AI provides clinical second opinions Doctors use AI daily as a clinical decision support tool to get second opinions without consulting colleagues or textbooks. It helps identify relevant trials, missing tests, and diagnostic possibilities. ## Drug discovery gets a boost AI is making the biggest impact in early-stage drug discovery by identifying promising molecules and biological targets. This is where AI excels at accelerating research that previously took months or years. ## The drug discovery challenge Researchers must match one of an infinite number of possible drug molecules to a disease target in human biology. Human biology has tens of thousands of genes, hundreds of thousands of proteins, and trillions of cells. ## AI identifies disease targets AI analyzes enormous amounts of data to rank potential disease targets that would have taken researchers months or years to identify. It creates a short list of the most likely candidates to attack. ## AI generates candidate molecules Once a target is identified, AI scours vast chemical spaces to find molecules that can disrupt that target. It can even help generate new molecules that don't yet exist. ## AI optimizes drug properties AI predicts which molecules will be absorbed by the body and reach the right tissue. It identifies the Goldilocks set of properties needed for a safe and effective drug. ## AI can be confidently wrong AI tools often express certainty even when they're completely wrong. This is a major concern in drug discovery where mistakes can be dangerous or wasteful. ## Humans must verify AI output Scientists and doctors must judge AI's output to determine what's promising and what's dangerous. They need wet labs, reasoning, and context to validate proposals and catch toxic compounds. > You still need a lot of judgment. ## Clinical trials remain analog Even with AI acceleration in early discovery, the second half of drug development remains slow. Researchers must recruit patients, conduct trials, and navigate regulatory approval. ## AI is an assistant, not a replacement AI augments human expertise rather than replacing doctors and scientists. The challenge is maintaining human agency and critical thinking skills as AI becomes more sophisticated. ## Healthcare needs disruption Dr. Khullar argues that healthcare is ripe for disruption and doesn't need frontier AI models to improve dramatically. Even older AI models could help significantly with current problems. > Healthcare is ripe for disruption and you don't even need the frontier frontier models to make it a lot better with AI. ## Don't shut down medical AI Rather than banning AI entirely, Khullar argues for slowing frontier models while allowing AI in biotechnology, scientific research, and clinical care. The benefits are too significant to abandon. ## Key takeaway AI is already saving lives in healthcare through administrative tasks, patient navigation, drug discovery, and clinical decision support, but it works best as a tool that augments human judgment rather than replaces it.