The Male Gaze

In the AI Story

The argument was not that every painter was conscious of imposing this structure, nor that every viewer was conscious of occupying it. The gaze was not in the eye of any individual. It was in the conventions of the medium — the invisible frame within which all visible choices were made. A painter could break the conventions, as Manet did with Olympia, and the breaking was legible precisely because the convention was so thoroughly established. Olympia looked back at the viewer in a way that refused the passivity the tradition demanded, and the refusal scandalized Paris.

Berger's analysis reached beyond painting. He argued that the structure of the gaze — the active male subject, the passive female object — shaped not only images but the relation of women to themselves. A woman was trained, from early childhood, to see herself as a spectator would see her. She watched herself walking, watched herself speaking, watched herself being watched. The gaze was interiorized, and the interiorization was the deepest form of the structure's power.

Laura Mulvey's 1975 essay 'Visual Pleasure and Narrative Cinema' extended the argument to Hollywood film and gave it the name — the male gaze — by which it is most often cited. Mulvey drew on psychoanalysis and film theory in ways Berger had not, but the structural argument was recognizably continuous. Both agreed the gaze was not individual but structural, not conscious but conventional, and not easily escaped because it was embedded in the medium itself.

The concept's relevance to AI is not a matter of AI systems literally looking at women. It is a matter of how the framework of structural gaze generalizes. When a technology is trained on a corpus that embeds specific assumptions about who is doing the seeing and whose perspective counts as default, those assumptions are reproduced in every output — not because any individual programmer chose to impose them but because the corpus carries them and the model extracts them. The gaze of the machine is Berger's framework extended to computational systems whose defaults encode specific cultural perspectives as though they were neutral views of the world.

Origin

Berger developed the argument across his art criticism of the 1960s, drawing on Kenneth Clark's 1956 The Nude — which Berger admired and fundamentally disagreed with. Where Clark treated the nude as a category of artistic achievement, Berger treated it as a site of political arrangement. The 1972 BBC series was the first full articulation. Its immediate effect in academic art history was mixed; its longer effect, traveling through feminist theory and cultural studies, was transformative.

Key Ideas

The gaze is structural, not individual. It lives in the conventions of a medium, not in any particular viewer's act of looking.

The conventions position a default viewer. In the European nude tradition, the default was male. In AI training corpora, the default is the culture of the internet's most prolific contributors.

The default is most powerful when invisible. A convention announced as a convention can be refused. A convention presented as the natural way of seeing cannot.

Breaking the convention requires recognizing it as one. Manet's Olympia was legible as a break because the convention was known. In AI, the conventions are not yet widely known, which is why breaking them is so difficult.

The gaze is interiorized. The most consequential effect is not what the gaze does to the image but what it does to the subjects of the image, who learn to see themselves through it.

Debates & Critiques

Feminist theorists have extended, revised, and complicated Berger's formulation for fifty years. bell hooks argued in 1992 that the male gaze as Berger and Mulvey described it was also a white gaze, and that the experience of Black women being looked at differed structurally from the experience Berger analyzed. Queer theorists have argued that the binary of male-viewer-female-object is itself part of the structure that needs dismantling. These extensions do not refute Berger's argument; they sharpen it. The general principle — that gazes are structural properties of media, not individual acts of looking — survives every refinement.