Generative Artificial Intelligence and Visual Communication: A Systematic Review of Transformations in Image Production, Audience Reception, and Meaning-Making in Digital Media

Over the last ten years, generative AI has been one of the most disruptive technologies in media and communications. The DALL-E, Midjourney, and Stable Diffusion tools allow one to create photorealistic pictures within a few seconds. They have transformed the economics of content production in visuals. However, they have also posed some basic questions concerning what visual communication is, the place of the sender, and meaning operations in digital media. Although this technology is rapidly growing and is increasingly being used in news, advertising and social networks, there is still no systematic review on visual communication literature in this field. The article will address this gap by conducting a systematic review of the literature published in 2018-2023. Web of science, Scopus, and Communication Abstracts were searched with a combination of keywords, which found 478 original studies. Following the application of inclusion and exclusion criteria, following the PRISMA protocol, 63 studies were finally chosen to be analyzed. These researches were summarized and discussed on three principal planes: first, the changes in the visual message production and the redefining of the concept of the sender in the communication models; second, the way audiences perceive, cognitively process and evaluate the credibility of the images generated by AI; and third, the changes in the media visual culture and in the new discourses of authenticity, reality and representation. The analysis is grounded in an integrative four-dimensional communication model—comprising semiotic, cognitive-affective, rhetorical-discursive, and contextual-technological dimensions—to evaluate the shift from indexical photographic evidence to interpretive algorithmic imagery. The results of this review demonstrate that generative AI has had three significant impacts on visual communication, as it: democratizes image production and eliminates the distinction between professional and non-professional creators; introduces a significant threat to the capacity of audiences to differentiate between real and synthetic images; and reproduces and reinforces existing cultural stereotypes based on historically biased data used to train algorithms. Another three key gaps found in the current literature highlighted by this review are inadequate attention to Western audiences, the preponderance of a quantitative methodology, and the lack of longitudinal research on long-term effects. On the basis of these findings, six research directions towards future studies are suggested.