The Monet Test: AI Prejudice, Attribution-Conditioned Formalism, and the Future of Aesthetic Judgment

A 2026 conceptual intervention by the artist SHL0MS presented a cropped reproduction of the 1915 Water Lilies by Claude Monet as a newly generated artificial intelligence image and invited viewers to explain its inferiority to an authentic Monet. Respondents supplied confident accounts of deficient composition, depth, color, intentionality, and soul, although the image reproduced an established Monet painting. This episode illustrates attribution-conditioned formalism, a process in which provenance organizes perception and formal vocabulary rationalizes the resulting judgment. This critical integrative review places the incident in dialogue with empirical aesthetics, creativity research, artworld theory, historical reattribution, and contemporary cases of mistaken AI detection. Controlled studies show that AI labels can depress ratings of beauty, profundity, creativity, skill, and monetary value even when stimuli are identical or difficult to distinguish. Conversely, blind evaluations have often favored AI-generated poetry, human-AI haiku, machine-produced humor, and AI-assisted stories, while large-scale research preserves human advantages at the highest levels of divergent creativity. Historical cases involving Marie Denise Villers and Judith Leyster demonstrate that gender and reputation have long altered the estimation of unchanged objects. The article argues that craft and quality remain relevant but cannot independently explain artistic value, which is mediated by provenance, institutional authority, presumed effort, social identity, and market prestige. It proposes a four-domain framework separating formal-perceptual, conceptual-interpretive, procedural-ethical, and institutional-social evaluation, together with a two-stage blind and disclosed review process. These reforms can preserve contextual interpretation while preventing provenance prejudice from masquerading as visual analysis.