From descriptive to reflective: Reading and interpreting generative AI Images
By: Seyedeh Maede Mirsonbol
| ARTICLE INFO: Volume: 11 Issue: 02:Winter 2025 ISSN: 2459-2943 DOI: 10.18680/hss.2025.0021 Pages: 47-68 Lic.: CC BY-NC-ND 4.0 |
KEYWORDS: Generativity Artificial Intelligence Semiotics Human-AI interaction Images |
ABSTRACT
As generative AI expands from text to image creation, critical questions emerge about the nature of meaning-making in machine-generated visuals. This paper theoretically explores the shift from ‘descriptive’ to ‘reflective’ reading of generative AI images, drawing on linguistic and semiotic theories of Chomsky, Halliday, and Culioli. Chomsky emphasizes formal, rule-based structures, while Halliday and Culioli highlight meaning as contextual and inferential. Although studies of AI-generated images often focus on surface-level features such as style, coherence, and resemblance, this paper argues for a semiotic engagement that considers the underlying structures and contextual processes (or their simulation) at play. The central question of whether generative AI engages in the complex, contextual processes of human meaning-making is explored through an applied approach focusing on prompting and image inquiry. This is to construct and encourage humans in showing reflection and interpretation of visuals, firstly, by describing the relationship between AI and humans to form an interactive partnership; secondly, by moving to a reflective interpretation that requires a human viewer to supplement AI’s syntactic fluency with socially, culturally, and cognitively grounded meaning-making.
Download full text: PDF
