A Bad Output That Reads Well: What I Told MDLinx About AI Risk in Medicine, and the Rest of the Answer
October 25, 2026 | Patient Safety, Plastic Surgery Consultation, Plastic Surgery News
When MDLinx asked physicians how worried they are about AI in medicine, I said moderately, and I said why: the failure I expect is a confident wrong answer delivered in fluent medical English to someone who cannot check it. The article ran one paragraph of that. Here is the rest: the patient who arrives alarmed by a diagnosis nobody made, the one told a procedure is simpler than it is, the three safeguards I would actually enforce, and why software needs the same post-market watch an implant gets.
