
Aug 28, 2026 · 18 min
Doctors adopt AI while questioning its authority
AI is everywhere in healthcare now. Doctors are conflicted
AI is entering clinical workflows faster than doctors, patients, and hospitals can agree on how to evaluate, govern, or take responsibility for it.
- 1Doctors use AI for documentation, research, diagnosis, imaging, and organizing patient information, but adoption often outpaces formal training.
- 2Medical chatbots can produce confident errors and hallucinated citations, making source-checking and critical evaluation essential to safe use.
- 3Privacy, liability, and clinical-reasoning concerns remain unresolved as hospitals and medical schools develop policies for rapidly changing tools.
Don't miss
Jonathan H. Chen compares AI’s convincing performance to magic, arguing that apparent authority and empathy can exist without genuine understanding.
The brief
Doctors are already using AI across documentation, research, diagnosis, imaging, and patient-information workflows, but the episode asks whether adoption is outrunning medical judgment.
Physician-computer scientist Jonathan H. Chen explains how medical tools, ambient scribes, and systems such as OpenEvidence differ from general chatbots—and why better sources do not eliminate hallucinations.
The central risk is not simply that AI makes mistakes; it is that overconfident answers, fabricated citations, and apparent empathy can make those mistakes harder to detect.
Chen warns that limited training leaves doctors and patients unprepared to evaluate AI, while hospitals face unresolved questions about privacy, malpractice, and rapidly changing institutional rules.
The standout comparison comes from Chen’s work as a magician: AI can appear authoritative or understanding through convincing performance, even when no genuine understanding is present.
The episode leaves clinical AI in a conditional place: useful when clinicians frame questions carefully and verify outputs, dangerous when they treat systems as authorities.
Featuring
Listen to the full episode and explore every guest, topic, and moment on PodLume.

Jonathan H. Chen
ChatGPT
OpenEvidence