ABSTRACT
“I just can't believe my eyes!” The expression of incredulity, “Believe it though I saw it myself, I cannot”, is a long-standing cliché. It appeared in Bartholomew Robinson's Latin and English Adages in 1621, indicating its use since at least the seventeenth century, and it became a cliché by the late nineteenth century. The recent and significant progress in artificial intelligence (AI), notably its capacity to produce highly realistic images, marks a pivotal point in the history of media and communications. The creative potential is breathtakingly exciting while simultaneously posing significant threats to public trust in visual media. The AI-driven image-synthesis progression continues at speed with ever-more sophisticated output, our ability to discern the authentic from the synthetic becomes increasingly difficult, leading to a widespread crisis of visual trust—challenging the long-established trust society has placed in visual media. The implications of this erosion of trust are significant and far-reaching as public perception, the integrity of journalistic output and even the collective understanding and memory of historical events are all on shifting sands. “I just can't believe my eyes!”. No, you probably can't… This chapter takes a look at current AI image generation tools and provides examples of just some of the instances where AI-generated visuals have been mistaken for genuine photographs and what the impact of those was on public perception. An important case study is the floods in Valencia, Spain in 2024 highlighting how questions of the authenticity of even real images can lead to a negative public response and subsequently have the potential to have critical issues such as climate change challenged. The psychological factors influencing how we perceive visual content in this era of generative AI are also explored including concepts of “truthiness” and confirmation bias. Where will the long-term consequences of a deterioration in trust in visual media, photojournalism, and collective-and-shared memory take us next?
