Critical Response IV: This Photo Does Not Exist: Generativity and the AI Gaze

For the past five decades of machine learning research in computer vision, the generation of synthetic images (or, AI photography) has achieved extraordinary success as a research area within the larger field of computer vision. Drawing on machine learning techniques of pattern recognition and the classification of visual objects, the astonishing gains of generative AI in producing synthetic, photorealistic images demonstrate that techniques of identification are just as easily turned toward fabrication. While the specter of deepfakes now threaten the integrity of democratic elections, these products of AI/ML are only the newest version of a problem coextensive with modern photography, whose speed of mechanical reproduction and dissemination has abetted “pseudo-events” and sociopolitical influence campaigns across the last two centuries.2 As synthetic image generation now reaches unprecedented levels of photorealism (as well as nonphotorealistic styles) and the industry has recently conquered realistic video generation at scale (OpenAI’s Sora, 2024), it is worth asking, as Amanda Wasielewski and others have done: how will this moment inflect the past, and future, of photography (see Amanda Wasielewski, “Unnatural Images: On AI-Generated Photographs,” Critical Inquiry 51 [Autumn 2024]: 1–29)?