There has been a lot of buzz about the emergence of AI technologies (artificial intelligence) over the past few years, but in the last six months, the buzz has grown far louder, at least in the community of artists I regularly circulate in. It's become loud enough that I decided I wanted to put forth my thoughts on the technology and maybe start a discussion about it.
AI Art vs Human-Made Art- What Is The Future, Really?
AI As An Aid to Artists
I first became aware of AI when several software companies, such as Topaz Labs and On1, began incorporating the technology into their photo editing software. Topaz, and a little while later, On1, brought out plugins for photoshop that helped improve image quality of photographs captured by humans. These plugins helped reduce digital noise in images, improve sharpness, and help bring out detail in images. None of this is anything crazy, or new, for that matter, but the addition of AI allowed the software to learn using different models, continuing to provide improved results and image quality.
Adobe has also gotten in on the act, with its newer neural filters for Photoshop. I've used them myself to colorize old black and white photos of my dad. The results are outstanding and it was fun to see my dad as a young boy in color, rather than the black and white images I'm used to seeing. You can see an example in the image below. In general, I have no issue with AI in these cases. The human element is still present in the art and the photograph that began the process remains at the end, captured by me- or in the case of the photo of my father, captured by my grandfather.
More recently, a number of text-to-image generators have appeared. I won't name them all, but one I've used is AI-Pro, and another is Google's Deep Dream. I've played with a few of them to varying results, none of which are particularly satisfying. But I say that with the knowledge that these artificial intelligence art generators WILL improve. Basically, the way it works is you enter some text, known as a prompt, to tell the software what you want it to create. I've shared some examples of my experiments below. I'll own up to the fact that I am new to this, and I know of several people whose results are astonishing. There's no denying that the AI can create whatever someone tells it to. But does that make it art?
Not in my mind. This is just my opinion, and I admit I may come off sounding like an old man yelling at some kids to get off his lawn, but art needs to have the human element to truly be art. I do not believe typing text into an AI generator meets that definition. Paintings need to have the canvas touched by a brush held by the artist. Photographs need to be captured by the photographer pressing the shutter button at the decisive moment. Be edited by the photographer. By contrast, while you can tell the AI what you'd like to see, it doesn't always follow directions. Maybe I just haven't gotten good enough at it yet, but try as I might, I couldn't make it generate a Maine lobster boat.
What Does It All Mean?
Ultimately, this is going to damage those of us who have chosen creative professions. Illustrators, landscape photographers, commercial artists, and writers (did I mention it could write too?) will all see their professions squeezed by those seeking to cut costs by using AI for everything from book illustrations and advertising photos, to entire novels. It's not there yet, but it doesn't take a genius to see where this is all headed. Do I feel threatened? Absolutely. While I have faith that many who appreciate my work will still see the value in human-created art, I know that others will settle for something AI-generated that looks "cool" and is much less expensive.
Ultimately, we'll have to wait and see where this all leads. Human-generated art has a certain quality to it that AI-generated art does not. It is often seen as a product of hard work, craftsmanship and creativity, allowing for a deeper connection between the artist, the artwork, and the audience. Human-generated art is generally unique and original, while AI-generated art is often repetitive and predictable in nature. Human-generated art seeks to capture true emotions, ties and events which cannot be replicated by AI-generated art.
The major shortcoming at the moment for AI-generated art, in addition to general lack of "realness", is the low resolution of the output. Human-generated photographs provide more meaningful and detailed information than AI-generated photographs. They are generally more dynamic and lifelike as they are taken in real-life situations, and can convey an emotional connection that AI-generated photographs cannot. Professional photographers, myself included often rely on human expertise and vision, rather than presets and algorithms, to create a more timeless, memorable and artistic photograph than AI-generated documents. In my experiments, I've been unable to replicate actual places with the AI. I'm sure that will change, unfortunately. The AI generators I've tried were unable to replicate such iconic places as Portland Head Lighthouse, and my attempts to render the Empire State Building resulted in an image of the building that lacked the famous spire at the top.
I should mention that my observations are limited to my own experience and what I have seen at large. I will note that recently in Australia, an AI-generated image won a photography contest, fooling the judges. However, to look at the image, I have to wonder how they were fooled. It's almost too perfect, but in the areas that are not perfect, the errors are obvious to me, as someone who has spent much of my life observing nature. The lower middle of the scene, where the water washes onto the beach on the left and right, yet is absent in the center, is not correct for the way water behaves when washing on a beach. The wave break itself is also unlike any breaking wave I've seen in nature. It does not look natural- specifically, the right angle of the crashing portion on the right. Water does not do right angles.
The coup de grace in that example is this quote from the "creators": “We didn’t need to wake up at sunrise, drive to the beach and send the drone up to capture the image. We created this image from our couch in Sydney by entering text into a computer program.”
Well, that's exactly WHY I go out and take photos in nature. TO WITNESS IT! Life on the couch can get awfully boring! To me, the creators of the software (not the image) have missed the point entirely. How sad.
Ultimately, there is no substitute for the human element in art and photography. The human emotion. The human vision. The human mistakes. I'm hopeful that most people will still find value in that, in the humanity of art.