For more than a century, photographs have carried a special epistemic authority: the idea that “seeing is believing” reflects the conviction that images could present evidence, document events, and settle disputes. Yet, with the rise of generative AI, images and videos can be fabricated with striking realism, which undermines the role of visual evidence in shaping our judgments of reality. This crisis is compounded by the spread of visual misinformation on social media, where visuals are often taken out of context or deliberately manipulated. Together, these forces contribute to the sense that we are entering a “post-truth” visual era in which images can no longer be trusted as reliable evidence.
This book resists the alarmist claim that public confidence in images is inevitably collapsing. Rather than framing AI as the end of visual trust, Visual Credibility in the Age of AI asks a more fundamental question: what has ever made an image credible? To answer this, the book draws on psychology, computer science, journalism, and communication research. It traces how images gained authority in domains such as visual journalism and science, and how notions of credibility have been constructed and contested within professional communities. It also unpacks the psychological mechanisms and cognitive shortcuts that shape how people judge what they see, such as fluency, vividness, aesthetic appeal, and contextual consistency. Finally, it examines the evolving technologies of image generation, their limitations and social impacts, and how audiences respond both to AI-produced visuals and to interventions aimed at preserving visual trust.
By tracing the historical, psychological, and technological roots of visual credibility, this book challenges the popular claim that AI will simply erase public trust in images. Instead, it argues that credibility has never been an inherent property of images but rather a practice negotiated among multiple stakeholders. Understanding this process clarifies how visual trust was established, why purely technical fixes are insufficient, and what steps in policy, media practice, and public education can foster more reliable ways of judging what we see.
This manuscript, Visual Credibility in the Age of AI, is currently under contract with Cambridge University Press and is expected to be released in late 2026 or early 2027.
YILANG PENG 