<p>Concerns about physician deskilling in AI-enabled medicine usually ask how automated systems substitute for, attenuate, or distort clinical capacities. This article argues that precision medicine introduces a distinct object-side problem. Its dominant classificatory operation is not simply finer nosological classification but stratification for action: patients are grouped and regrouped for treatment selection, risk allocation, surveillance, trial eligibility, and prediction. In doing so, precision medicine fragments established disease categories, connects conditions across former boundaries, temporalizes disease into trajectories and thresholds, and produces actionable risk states and algorithmic subgroups. If clinical expertise is acquired through relatively stable nosological categories, then preserved skills may become insufficient when those categories no longer organize the clinical object in the same way. I call this failure of fit ontological deskilling. The article distinguishes it from technological deskilling, diagnostic uncertainty, overdiagnosis, medicalization, plastic diagnostics, and ordinary adaptive-learning demands. Drawing on Canguilhem, Mol, phenomenological intentionality, and accounts of expertise, it argues that ontological deskilling changes the conditions of clinical judgment, uncertainty, and responsibility. The appropriate response is critical nosological awareness: the ability to use disease categories competently while recognizing their contingency, plurality of functions, and susceptibility to transformation.</p>