Algorithmic Gaze in AI-Generated Digital Portraiture

AI-generated digital portraiture has become a common artistic medium, but the representational conditions behind its outputs remain insufficiently examined in art research. This study investigates how prompt language organizes visible identity before an image is produced. Based on selected literature on algorithmic bias, AI ethics, text-to-image stereotypes, and visual culture, the paper develops a sixteen-item prompt corpus for reading digital portraits. The analysis proposes four audit categories: default visibility, facial legibility, cultural compression, and aesthetic normalization. The article argues that a generated portrait should not be judged only by technical finish, because the short prompt can make some identities appear ordinary while making others require specification, correction, or cultural marking. The study further argues that algorithmic gaze enters portrait production through prompt language and affects the artistic reading of identity, culture, and style.