We examine the tension between professional expertise and acceptance of algorithmic advice among knowledge workers. Drawing on the cognitive entrenchment perspective, we argue that domain experience fosters highly efficient yet cognitively entrenched schemas that may constrain receptivity to external algorithmic input. Using data from an AI-enabled prescription monitoring system that flags medical prescribing risks such as overdosage, underdosage, and drug interaction under random patient–physician assignment, we find that experienced physicians are significantly less likely to follow AI recommendations. This experience-induced resistance is particularly pronounced among physicians with low openness to experience, high neuroticism, greater work fatigue, and more complex patient cases. Further, adherence to AI warnings significantly reduces patient revisit rates, ruling out inferior algorithm quality as an explanation for experience-driven AI resistance. These patterns suggest that AI resistance among experts emerges not from low algorithm quality but from expertise-induced cognitive rigidity that alters trust calibration toward algorithmic systems.