Art Papier, M.D., issued a direct warning to dermatologists at Fall Clinical Derm 2026: do not assume artificial intelligence is always right. Managed Healthcare Executive reported the session, in which Papier — a longtime voice on clinical decision support — pushed back against the growing habit of treating algorithmic output as ground truth.

The venue matters. Fall Clinical Derm is one of the field's largest continuing-education meetings, and its 2026 edition put AI squarely on the agenda. That a speaker of Papier's standing used that platform to caution rather than celebrate is itself the story.

What's the core of the warning?

Papier's message, as covered by Managed Healthcare Executive, is simple: AI systems can be wrong, and clinicians who treat their outputs as infallible are accepting risk they may not see. The warning lands at a moment when dermatology is one of medicine's most AI-saturated specialties — diagnostic triage tools, image classifiers and documentation assistants are already in the workflow.

The caution isn't that AI lacks value. It's that value collapses when users stop interrogating the output. A tool that's right most of the time is most dangerous precisely at those moments when it isn't, if the human in the loop has stopped checking.

Why does this matter for dermatology specifically?

Dermatology has been an early and heavy adopter of machine learning, particularly for image-based diagnosis. That makes the specialty a natural test case for what happens when confidence in algorithms outpaces the evidence behind them. Papier's warning reads as a counterweight: the clinician's judgment remains the safety layer, and that layer only functions when it stays active.

What does this change for practicing dermatologists? Not the tools themselves — the behavior around them. The practical takeaway from the session is to keep verifying AI-generated outputs, whether they're diagnostic suggestions, documentation or decision support, rather than defaulting to agreement.

The bigger picture

Sessions like this one signal where the profession's internal conversation is heading: from "can AI do this?" to "how do we supervise it?" Expect more meeting programming on AI verification, error modes and clinician accountability as adoption deepens — and expect speakers like Papier to keep pressing the point that the technology is only as safe as the skepticism of the person using it.