Featured

The Efficacy of Images in the Age of AI: An Ontological Inquiry into Visual Tension

Published: 20 June 2026, Zenodo. DOI: 10.5281/zenodo.20770298.

Zenodo record: https://doi.org/10.5281/zenodo.20770298

Abstract

AI image-making has made content and style easier to adjust, yet it does not provide sufficient grounds for judging the efficacy of images. This article defines image efficacy as an image's capacity, in presentation and viewing, to affect perception, guide looking, and support the unfolding of understanding, and uses that definition to examine the relational status of visual tension. Through conceptual analysis and comparisons of process materials for Guernica and of AI images, it argues that certain visual effects are determined only within shared relations, and proposes a comparative test for whether pull and constraint constitute conditions of organization. Relational necessity, tension's conditional role in organization, and the ontological claim that tension is prior to the subject-object divide are treated separately. Visual tension earns its explanatory value by bringing relative weight and directed effect into image comparison, making the grounds for selection and revision specifiable and testable.

Keywords: image efficacy; visual tension; relational ontology; AI images; TSG.

This is the foundational first statement of the TSG (Tension–Structure–Generation) theory of image generation, published to establish priority. The full-length study is in progress:

《TSG艺术生成主义》／TSG Art Generativism — book-length study in progress.