While much discussion is focused on the relation between generative artificial intelligence and visual culture, its implications for journalism and public perception require a more in-depth critical examination. This study examines 28 AI-generated images created using DALL-E and DeepAI tools by addressing the ways artificial intelligence constructs visual narratives of the 2023 Israel-Hamas conflict. By employing a visual discourse analysis combined with social actor theory, this essay explores how civilians are represented, categorized, and positioned within digitally rendered conflict imagery. The study highlights how AI-generated imagery reproduces dominant tropes from traditional war photography while simultaneously reshaping them within new technological frameworks. The analysis reveals that the AI-generated visuals consistently foreground the civilian experience, particularly focusing on themes of destruction, resilience, motherhood, and childhood, while omitting direct representations of the primary political actors involved in the conflict. This pattern likely reflects the characteristics of the data used to train contemporary generative AI systems, including biases toward humanitarian imagery and constraints that discourage the depiction of identifiable political actors, thereby privileging emotionally resonant civilian narratives over politically explicit representations.
Theodosiou et al. (Mon,) studied this question.