More on AI art vs. photography. I find it interesting that some of the same arguments were originally used to denigrate capturing images on film. "It can't be art, because all you're doing is pushing a button."
Having been an avid photographer since I was about 9 years old, I am still very much an amateur. I take my hat off to people who make photos for a living, and who have mastered the craft.
We can all tell the difference between a great photo and a snapshot. Granted it's gotten much easier to make great photos. Digital photography is much more forgiving than film. Cameras are now sophisticated computers. Not only can they do perfect exposures, but they can process the image live using HDR to preserve detail in both light and shadow that would have been impossible on film.
Photographers do other tricks beyond even automated digital processing, like bracketing and stacking. There are add-ons like Arsenal, that further optimize what a digital camera is capable of. You can set a camera on a tripod in front of a building or busy street, and the system will keep taking photos until it's captured the entire scene enough times that people can be removed from the shot, leaving only the architectural subject. It does this automatically. Cheating?
No matter how sophisticated the capturing technique, photography is still by every measure, art. Also, politics. Also, deception. Neil Postman, the visionary media critic who wrote "Amusing Ourselves to Death," had a lot to say about this. Photography seems like truth-telling, since the film merely records the photons that enter the camera. But if you took that literally, you would have to believe that photojournalism could not lie. Postman's point is that photography cannot help but lie. And that a photograph is never objective reality. Scientifically, film is exposed only by the photons that strike the emulsion. Indeed the film captures whatever is there.
But here's the caveat: THE FRAME.
By putting a frame around reality, we are editorializing. It's like the observer effect in physics, which states that a particle cannot be observed (measured) without changing the state of that particle. The presence of the photographer alters the scene. Anyone who photographs human or animal subjects knows the biggest challenge is to capture truly candid behavior.
The photographic term "framing" has entered the lexicon of political strategy. In the sense that persuasion entirely depends on how a question is asked. This has deep roots in social science. Are you "pro-life," or are you "pro-forced-birth?" Same question, different framing.
Much of recent history is told in photographs. But those images, by definition, exclude whatever is outside the frame. In many cases, things outside the frame are superfluous. But that was the photographer's "snap judgment," was it not? Sensibilities change in different eras, and what a given photographer thought was important, might be completely different than what we might find important today. Think of the images not taken by photographers pursuing assignments based on the "great man" theory of history. For every photo of a president or public figure, there are hundreds of photos not taken of their surroundings that would be more historically revealing than the images we see.
(Too bad frame-eliminating 360Β° VR cameras didn't exist over the last century--imagine what we would have seen, including the photographer of every historic image.)
Fortunately, the sheer volume of photos beginning in the late 19th century, give us a window on what we might have missed. There's always been a counterculture of photographers interested in the ordinary. And I would argue those images are of great, lasting value.
Then there are giants of photography like Ansel Adams, who was born at the turn of the 20th century, and gave us stark images of nature, captured in a way no one had quite seen. No one had captured the framing or contrast as he did, stripping natural features down to their absolute essential visual nature--*from a specific point of view.*
To look at one of his photographs is to experience something that didn't exist before he made it.
Late in the 20th century, the "reality" of image capture by photography was further blurred by airbrushing, and later Photoshop. Inexorably, art, photography, design, and visual effects have melded into one grand discipline. Seeing is most definitely no longer believing. Pixels are reality-distorters and cannot ever compose Reality.
Enter AI art.
Generative AI is another level of computer-assisted reality distortion. With a palette trained on billions of images, a user of these tools has full command of every photograph, painting, and film in human history, and can dial up anything they want by describing it with text. This has, predictably, led to an amazing explosion of outrage and bad thinking. Some artists and photographers are terrified they will lose their livelihoods. Others embrace the tools. Still others see outright theft, or at least a devaluation of craft and skill.
This is all, to put it bluntly, shrill nonsense.
There are certainly questions about ownership, and compensation of artists, and loss of revenue by illustrators and designers. And these tools can sometimes come close to plagiarizing original images. If you ask for a specific image of Batman or The Joker from a specific scene from a specific film, an AI art generator will render something very close to that scene. And that does raise intellectual property questions. Eventually this will all be sorted out, in the same way that all questions of copyright and fair use have been adjudicated. If some kid wants to render a scene of The Joker eating Cheerios at a breakfast table, does that really hurt anyone? If that kid wants to publish that scene and profit from it, without crediting and compensating the original IP owner, that's a different matter.
I'd like to point out that Mickey Mouse isn't that difficult to draw, and yet long before the invention of AI art tools, people were prevented by Disney's deep-pocketed lawyers from using the character in any commercial venture whatsoever. And yet, kids could still doodle Mickey Mouse to their heart's content. After 100 years, the earliest versions of that character have now become public domain. There's a process in place for IP protection, and that won't change as a result of AI tools.
Now let's tackle the question of ease of use, or "button pushing." It's trivially simple to come up with mediocre images using Midjourney or Dall-E. Our social media feeds are filled with them. There are "photographs" of people who don't exist, endless comedic distortions of public figures, things that never happened, and non-stop fake images of nature. At this point, "real or AI?" isn't a question most people can reliably answer. Distorted and "impossible" images go viral just like other memetic lies. That game is over, and Reality has decisively lost.
But what of making actual art? I define art as the skill of conveying an emotional message in any media. Just as putting a frame around a scene is a part of the art of photography, selecting topics and juxtapositions and styles is the art of AI. And lots of frustration, trial and error, and worthless results.
Yes, I can command the machine to access any of billions of photos or prior art. No I didn't make those images. No, I don't have the skill with a pencil or paint brush to draw them myself. But what I can do is select specific elements, for a specific purpose, to convey a specific emotion, and use the machine to put them together in a way no one else has.
Creating AI art is not remotely just about writing good prompts. Like photography, it's recognizing a good scene when you see it. Like design, it's knowing what represents good composition. It's curation, rejecting 97% or more of what the machine spits out. It's modification and editing. It's frustration when you want to change one small element, and the machine spits out a completely different image. It's starting over when modification isn't working. It's learning to work with software revisions that completely change how the system responds to prompts.
Communicating with an AI is not unlike communicating with human artists and designers. Where you have a concept in your head, and you try to describe it to that person in words, and they think you mean something completely different, interpreting it according to the concept they have in *their* head.
In spite of all the controversy, we are witnessing the messy birth of a new artistic medium. Eventually, these kinks will all be worked out. And what that means is unfettered artistic expression of whatever you can imagine. Some think that just means an endless parade of unartistic slop. But that's true of any medium. As long as there have been pencils and paper, there have been bad drawings. Yet we recognize the good and the great.
I'm as much of an amateur at AI art as I am an amateur photographer. But that doesn't mean I can't do good work, and derive artistic fulfillment from it, and maybe even some compensation. Because in the end, if I create a pleasing image that someone wants to look at, that conveys some emotion or meaning they enjoy, isn't that adding value that didn't exist before?
Given the permutations possible, it's unlikely that anyone will duplicate my work, or that I will duplicate anyone else's. And that is why I know that AI artwork by people who are actual masters at that craft, will eventually take its place among fine art. Eventually art schools will incorporate AI tools along with pencils and paint. And the resulting work will speak for itself.
Same goes for music, same goes for writing, same goes for filmmaking. In the hands of people who want to convey their own unique ideas, eventually these AI tools will become simply, art supplies.
#aiart #art #photography
