Turn on cookies to sign in
Signing in saves your progress to your Curie account. We can only do that with cookies on — turn them on to continue.

AI is increasingly embedded in clinical workflows — but what happens when it contradicts a clinician's judgment? Emergency physician Harvey Castro, MD, says the answer is clear: the clinician owns the decision. At the AIMed26 conference, he outlined how providers should document disagreements, report AI errors, and guard against "deskilling" — the risk that over-reliance on AI erodes clinical judgment over time.
AI tools in healthcare are no longer just offering suggestions — they're taking actions inside hospital systems. At the AIMed26 conference in Orlando, emergency physician Harvey Castro, MD, MBA, chaired a panel on what happens when AI and clinicians disagree, and his message was direct: the physician keeps the moral compass, the machine finds the patterns.
Castro emphasized that clinicians should treat AI like "a smart colleague who has never met your patient" — useful, but always worth checking against the person in front of you. When disagreement arises, he recommends documenting what the AI suggested, why you disagreed, and what you did instead. He also stressed that hospitals must create clear channels for reporting AI errors, treating them like any other clinical safety issue.
Key Takeaways:
Why it matters: As AI becomes a permanent fixture in clinical care, the stakes of getting the human-AI dynamic wrong are high. Castro's framework — human plus AI, with the physician retaining accountability — offers a practical roadmap for clinicians navigating this new reality.