The Least

A patient's father received an open abdominal operation following an infection, as the hospital lacked the equipment for interventional radiology (IR). The surgeon proceeded with the open operation based on the available options. The author suggests that a family member could have obtained another option by consulting a physician, which was possible in the father's case. The author contrasts this historical experience where patients were limited by the choices presented to them with the current reality where patients have access to artificial intelligence (AI) for seeking alternatives. While this offers patients more options, it also introduces risks of misinformation, as demonstrated by AI failing to produce accurate initial diagnoses 80% of the time when given only symptoms. Furthermore, the adoption of AI by patients is outpacing that of physicians; while physicians use AI professionally, their use for clinical diagnosis remains limited (only 13% utilize it), leaving the physician at risk of being less informed than the patient interacting with an external AI source.

The narrative shifts from a personal anecdote about patient advocacy to a structural critique of the evolving physician-patient dynamic in the age of AI. The core tension lies between democratized information access (via consumer AI) and the established necessity for expert clinical judgment, suggesting a potential systemic erosion of the physician's epistemic authority. The failure of general-purpose AI to provide accurate diagnoses highlights the critical difference between AI reasoning based on patient narrative and the higher accuracy achieved when AI processes the full medical record alongside a clinician’s review—a distinction rooted in context. This prompts an analysis of "cognitive spoofing," where the fluency of AI can mask a lack of genuine clinical understanding, suggesting that the burden shifts from obtaining information to verifying competence. The proposed solutions—equipping physicians with clinical AI, encouraging their use of consumer AI to understand failure modes, and establishing robust accountability guardrails—suggest an acknowledgment that technological integration must serve to reinforce, rather than undermine, professional responsibility. The implication is that the necessary evolution involves ensuring that the physician remains the ultimate locus of informed decision-making by integrating new tools while strictly maintaining professional boundaries and oversight.