AI Resilience Report forGeneral Internal Medicine Physicians

Right now, AI in internal medicine is mostly augmenting doctors, not replacing them. The biggest wins are on the paperwork side of the job—those first two tasks about writing reports and maintaining patient records. Ambient AI "scribes" listen during a visit and draft the clinical note automatically, and health systems are rolling them out to cut administrative workloads and free physicians for patient care ^[1].

Adoption is climbing fast: an American Medical Association survey found 81% of physicians now use AI professionally, with an average of 2.3 use cases each ^[2], and that figure has more than doubled since 2023 ^[3]. AI is also nudging into diagnosis and prevention—community health centers are piloting AI clinical decision support to flag risks and suggest next steps ^[4]—but the actual treating, immunizing, and long-term care of patients still relies on a human doctor's judgment and relationship with the patient.

Adoption is moving quickly for documentation tools because they're commercially available, plug into existing electronic health records, and pay for themselves by reducing burnout and giving doctors more billable visit time. But adoption is slower for higher-stakes clinical decisions. In a Doximity survey, more than 70% of physicians named accuracy and reliability as their top barrier to using AI ^[5], and a study of AI-assisted virtual urgent care visits showed physicians frequently changed the AI's initial recommendations ^[6], underscoring that human oversight is still essential.

Professional groups reinforce this: the American College of Physicians has formally recommended that AI augment—not replace—physician decision-making ^[6]. The takeaway for you: the human parts of medicine—listening, judgment, ethics, and trust—are exactly what remain valuable.