<p>The rapid integration of artificial intelligence (AI) in ophthalmology has produced remarkable diagnostic accuracy while simultaneously raising significant ethical and epistemic concerns. This article argues that the prevailing AI paradigm fosters a form of data-fetishism: the tendency to treat datasets and algorithmic outputs as objective truth, privileging measurable metrics over the patient’s lived experience and clinical context. Inspired by Marx’s concept of commodity fetishism, the analysis demonstrates how this phenomenon produces “algorithmic substitution”: a process through which the clinical encounter is converted from an ethically engaged encounter into a data-centric transaction, subordinating beneficence to statistical optimization. Four interconnected manifestations are examined: (1) dataset bias, where the marginalization of rare diseases and non-Western populations constitutes an epistemic injustice; (2) citation bias, which reinforces academic power structures; (3) the “covidization” of research, illustrating metric distortion; and (4) language hegemony, resulting in the epistemic erasure of regional medical knowledge. Clinical evidence reveals that these biases produce systematic errors in high-stakes domains such as ocular oncology and retinal surgery. The analysis concludes that data-fetishism enables a “moral deskilling” of ophthalmologists, redirecting professional fidelity from the suffering patient to the optimized dataset. Preserving the fiduciary essence of medicine requires a reorientation towards epistemic justice, emphasizing representativeness, ethical design, and the maintenance of the clinical gaze.</p>