Photography after AI

A paradox: 'the photographic' is central to visual culture inflected by artificial intelligence (AI), and yet AI is rapidly destabilising photographic practice. A cyclical theme, the notion of what constitutes a photograph and what it means to be a photographer are fundamentally questioned once more, especially since the release last year of novel text-to-image models capable of generating so‑called photorealistic images from text prompts (DALL·E, Stable Diffusion, Midjourney, etc.). In a curious reversal of the conventional process of captioning a photograph after it has been taken or printed, professional image makers and publishers are now negotiating the ability to synthesise photographic-looking images from executable texts. In the process, the still-dominant paradigm of photography as a contingent encounter between a camera user and the world is being upended by proprietary AI algorithms trained on datasets of existing photographic images, raising vital ethical questions on authenticity, biases and authorship.