When Good Intentions Aren’t Enough: AI, Medicine, and the Risk of Losing What Matters Most
Artificial Intelligence is transforming medicine at a breathtaking pace. With nearly 700 FDA-approved AI devices already in clinical use, the promise of greater diagnostic accuracy, improved efficiency, and reduced costs is real and compelling. As physicians, we should embrace tools that improve patient care. But we must also ask a harder question: at what cost?
In a commentary published in Otolaryngology–Head and Neck Surgery (see attached), I argue that AI’s most significant unintended consequence may be the erosion of the physician-patient relationship — the very soul of medicine.1 This concern is not abstract. U.S. Surgeon General Vivek Murthy has already declared a national epidemic of loneliness and isolation, and the evidence suggests that increased reliance on technology — even well-intentioned technology — contributes to that isolation.
We have seen this before. The electronic medical record was designed to improve care but in practice pulled physicians’ attention away from their patients and toward their screens. AI risks doing the same, at greater scale and speed.
Some argue that AI can actually improve relationships — AI responses to patient questions have even been rated as more empathetic than those of real physicians. But empathy without authentic human experience is a simulation. When AI says “I am sorry to hear that,” the statement is grammatically correct but fundamentally hollow. Authentic relationship requires two people.
The path forward requires vigilance on two fronts: physicians must intentionally protect human connection in every clinical encounter, treating AI as a tool that frees them to be more present — not less. And AI developers and policymakers must establish binding bioethical principles that prioritize the physician-patient relationship over efficiency and profit.
The road paved with good intentions is still a road that can lead somewhere we do not want to go.
Reading this made me think back to the summer I spent volunteering at a community health clinic that was just starting to experiment with an AI‑driven triage system. The software was impressive—quickly flagging high‑risk patients and suggesting treatment pathways. Yet, I watched a nurse pause before delivering a diagnosis because the AI’s recommendation didn’t align with the patient’s story. She asked the patient about their recent stress at work and family dynamics, things the algorithm could never weigh. melbet In the end, they adjusted the plan based on that conversation, and the patient recovered faster than the data‑driven model had predicted. That experience showed me how easy it is to get swept up in the promise of technology and forget…