Adobe is going all in on Artificial Intelligence (AI) in their products, most notably for us, Lightroom and Photoshop. I’ll admit to watching, somewhat slack-jawed as generative fill seamlessly removes distracting elements from an image. Other than some concern this will encourage laziness behind the camera, I am all in on this and other related applications of the technology. I think A.I. offers great promise in assisting with many of the mundane, but no less important tasks, such as categorizing and keywording images using image or object recognition, or automated “first-pass” culling of large sets of images from a completed job. Wedding and event photographers may find this helpful in selecting images with eye blinks, subject motion, or imperfect focus, presenting them for final review and rejection.
I have seriously mixed feelings, however, on the wholesale use of generative AI in photographic projects to create, or alter important image elements, not the least of which are the ethical implications for responsible journalism vs. the spread of misinformation. But beyond those serious issues, I think overusing or even using it at all transforms photography as we have come to know it into something decidedly un-photographic.
Photography, as with most other art forms, brings together two skill sets: the vision of the artist and the physical skills to express that vision outwardly. Painters and sculptors possess or develop a level of hand-eye coordination most of us can never hope to achieve — allowing them to express on canvas or in stone the artistic vision in their mind and imagination. Similarly, photography demands technical knowledge and proficiency in the physical skills needed to wield the tools — cameras, lenses and software — to express something about their subject that goes beyond capturing a mere likeness.
“Art” is arguably a uniquely human creation. As far as we know, humans are the only animals capable of abstract thought. Abstract thinking involves absorbing information and making connections within the wider world, along with understanding concepts not part of the physical world such as freedom, loneliness, or humour for example. Understanding the concept of symbolism also arises from this ability for abstract thought. Abstract thinking is a necessary precursor to creating art. Without the ability to form connections or recognize symbolism, photography becomes mere record-keeping.
Certain animals indeed produce creations that some have suggested constitute “art” — the Bower bird creates elaborate arrangements of objects around its bower, and the Pufferfish creates intricate circular patterns in the sand on the sea floor. However, it is widely agreed that these displays are directed at attracting a mate rather than personal expression. Human art is not simply about perpetuating the species.
Photography, compared to other forms of visual expression is largely observational. Photographers observe the world, find meaning there and convey that meaning to their viewers. Compared to painting, photographers can only deal with what is in front of them. The challenge is finding meaning within what is in front of the camera at the moment of exposure. This forces the photographer to deal with distracting elements or inappropriate light in-camera before the image is created. Certainly, digital post-production tools allow photographers to deal with these issues to a greater or lesser extent after the image is created, but for most of its history photographic processes demanded that these issues be dealt with in-camera. Painters on the other hand could simply elect not to paint the power line running through their landscape.
Introducing generative AI alters the photographer’s connection between vision and expression. While some argue that AI enhances creativity by overcoming technological hurdles between imagination and expression, AI inevitably influences the expression of any idea or subject it is asked to create by inserting into the result its “own” biases arising from whatever data set it was trained on. So, I think it is rare that a digital artist creating original work using AI tools can claim full artistic ownership of a work created in this way. Consider…how much artistry is involved in simply telling an AI image generator to, “Create a dark low-key chiaroscuro portrait of an older 16th-century Italian gentleman in the style of Caravaggio”, to produce the lead image in this article.
Even though photography is still largely observational, nothing in that idea precludes non-representational work (often referred to as “abstract”) since even these images are based on deep powers of observation and recognition. Nor does its observational foundation preclude the idea of composites created with multiple images. Here again, abstract often surreal ideas come to life through observation, forming connections and recognizing symbolism. Ryszard Horowitz, Jerry Uelsman and John Paul Caponigro all produce stunning, often surreal imagery using photographic composting techniques.
A personal favourite, Horowitz’s images have sometimes been compared to the work of Rene Magritte and Salvador Dali. He creates his images using purely photographic techniques. Working initially on film and later digitally, he creates astounding images suggesting imagined dreamscapes interlaced with symbolic imagery. A favourite of mine, his image, “Allegory” (reproduced below) is made up of a half-dozen or so images, all shot on film, later scanned and composited digitally.
Looking at, and appreciating both the raw skill and the exceptional artistic vision of Horowitz leaves Generative AI a distant “also ran” in my opinion. So, for me at least, I can’t get behind the idea of allowing an AI image generator to create all or any significant part of an image that will be called “photographic”.
It may legitimately be called something else, it may even be considered Art, just don’t call it Photography.