Abstract
This paper proposes a method for critically approaching GAN-generated images through the tools of literary and cultural criticism. Taking Trevor Paglen’s “Adversarially Evolved Hallucinations” (2017-ongoing) as a key example, the study focuses on the visual indeterminacy produced by generative adversarial networks and argues for an interpretive model rooted in what might be termed “apophenic reading” (the projection of meaning onto opaque or ambiguous visual data). The concept of the flatline construct, developed by Mark Fisher, serves here as a central aesthetic category: a state in which affect, perception and representation collapse into a frictionless yet ontologically unstable surface. Building on Fisher’s theorization, the paper reads GAN images not as flawed simulations but as artifacts that demand new descriptive and interpretive practices. Drawing on narrative motifs and affective categories from literary criticism, such as the uncanny, the eerie and the weird, the paper refines existing theories of AI visuality while situating them within a broader cultural logic. It also engages Jacques Rancière’s work on image indeterminacy and the dialectics of passivity and activity in spectatorship. By proposing an updated theory of ekphrasis, aligned with Hannes Bajohr’s “operative ekphrasis” and Mieke Bal’s “travelling concepts”, the study offers a framework for critically describing AI-generated images beyond conventional aesthetic vocabularies. It argues that GAN imagery challenges both the ontology of the image and the epistemology of its reading, calling for a hybrid method that reclaims literary tools for visual culture analysis.
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
2025 Special Focus—From Democratic Aesthetics to Digital Culture
KEYWORDS
GAN AESTHETICS, FLATLINE CONSTRUCT, APOPHENIC READING, LITERARY CRITICISM, EKPHRASIS
