AI Can Imitate Nature. It Cannot Witness It.

Generative AI can already produce spectacular wildlife images: perfect light, dramatic behaviour and animals that never existed. As the technology advances, distinguishing such images from photographs will become harder.

But visual perfection is not the essence of nature photography. Its defining quality is the relationship between an image and the event it records.

A true nature photograph begins with an encounter. The photographer must find the location, understand the species, read the conditions and wait, knowing nature may not cooperate. The resulting image records something that happened at a particular place and moment.

That connection to reality is more than a matter of taste. Leading institutions still define photography through its relationship with the world. The World Press Photo Contest requires photographs to be made with a camera and excludes synthetic images and generative fill. Nature Photographer of the Year likewise prohibits AI-generated and computer-rendered images.

The importance of encounter becomes even clearer underwater, where making an image demands photographic judgement, diving skill, restraint and environmental awareness.

Underwater, the encounter matters even more

Underwater photography requires the photographer to manage buoyancy, currents, limited visibility and shifting light while working with an animal free to disappear at any moment.

Ethics are inseparable from that process. Sometimes the most important decision is not to take the photograph: never touch the reef, manipulate an animal or pursue a subject showing signs of stress. The British Society of Underwater Photographers’ Code of Conduct places the welfare of marine life and the environment above obtaining an image.

AI experiences none of this. It does not wait in cold water, control its breathing or retreat from a vulnerable animal. It generates a plausible scene without witnessing one. In underwater photography, where the image reflects choices made before the shutter is pressed, that absence matters.

For the same reason, underwater competitions treat synthetic imagery as more than a technical variation. The Underwater Photographer of the Year competition strictly forbids AI-generated images and image content, and verifies shortlisted entries using their RAW files. That process connects the finished image to its original capture and the encounter it represents.

AI is a tool, not a witness

This does not make AI incompatible with photography. Used transparently, it can assist with cataloguing, research, captioning and limited image processing. The boundary is crossed when it invents an animal, changes behaviour or constructs a scene. Then the result is no longer documentary nature photography. It is digital illustration.

As synthetic imagery becomes more convincing, the question will be not only what an image shows, but how it came into existence. Provenance will therefore matter more. Initiatives such as the C2PA Content Credentials standard are developing verifiable records of how digital content was created and edited.

Such records cannot recreate the experience of being in the water. They can, however, help establish whether an image began as a genuine capture and what happened afterward. Provenance thus reinforces the ethical and documentary standards that distinguish photography from generated imagery.

AI may replace some demand for inexpensive, decorative images of nature. It may also become an important artistic medium in its own right. But it cannot replace the defining value of nature photography: "Someone was there. Something happened. And the photograph bears witness."