Three doctors walk into a room to talk about artificial intelligence, and it turns out two of them built the AI and all three have patients to see. Somewhere a hospital administrator is nervously refreshing their inbox.
DOC is convening a conversation on medical AI, and unlike most sessions on the topic, nobody in this one is speaking from the sidelines.
Jordan Shlain, M.D., built Private Medical after a career watching a strange chart play out in real time: the number of hospital administrators climbing steadily for forty years while the number of doctors barely moved. He has a theory about why health systems buy new technology, drawn from his own years selling software into hospitals. There are four reasons, in order of how often they apply: A regulation demands it. It limits legal exposure. Everyone else is doing it and nobody wants to be the one who wasn’t. And, bringing up the rear, it occasionally helps the patient.
Travis Zack, M.D., Ph.D., leads the clinical side of OpenEvidence, the tool a lot of doctors now open before they open anything else. His worry is not that AI will get a diagnosis wrong. It’s that most of the AI being built for health care was never asked to think about the patient at all.
Byron Crowe, M.D., works the structural end of the same problem: getting AI systems past the demo and into real use, selling direct to patients and to enterprise health partners in a world where a bad output has a name attached to it. He has a meeting with a state medical board the morning after this conversation, which tells you where his head actually lives. His concern is structural. The industry knows how to build these tools now. It has no real answer for who gets to approve them, audit them, or take them away.
Where the three of them agree is more interesting than where they’d disagree on, because it’s not really disagreement, it’s the same worry examined from three separate desks. Someone, somewhere, is going to use the argument that AI is safer than a doctor to quietly take the decision out of the doctor’s hands altogether. Whether that’s a scandal or a step forward may depend entirely on who’s asking.
Jordan puts it simply. Forget who wants to stay in control. Start with what’s good for the patient, and work backward from there. It sounds obvious until you notice how rarely anyone in the room actually starts there.
This is not a panel about whether AI is coming for medicine. It’s already here, quietly filing claims and half-writing your doctor’s notes. The real question, and the one this group is uniquely equipped to answer honestly, is who ends up holding the pen when it matters.
Join us at DOC 2026 this year to be in the room when this conversation, and many more like it, happens live.