AI-generated images of “cats” offer novel opportunities to consider the semic role of expectations in sign formation where they act as a constraints on semiosis through the potential identification of the AImage as “correct,” or as a “glitch.” Because the identification of “errors” depends on a range of technical and cultural expertise, they offer valuable insights into the interpretive process. The automated generation of media by AI separates the artist’s decision-making process from image production, continuing a trajectory that began with (...) the invention of photography in the nineteenth century that brings the sign formation process into consciousness by distinguishing “intentional” and “unintentional” encoding. The identification AI-produced images as-glitched provides a vehicle to consider how the sign formation process informs identifications of creative action as an intentional action: aesthetic appraisals are central to this process where cultural beliefs about creativity become ideological constraints on interpretation. The potential to understand AI “glitches” as expressive features of the image-object, rather than errors, proceeds via the aesthetics and affects of earlier art, such as the “painterly motion” shown in old master paintings by Peter Paul Rubens, or via the heritage of Surrealism. These affective constraints on sign formation reveal the central role of “glitches” in the distinction of creative and uncreative action. (shrink)