I don't come here a lot anymore.
But when I do, I do so to see what's getting produced by the community. "What's popular today? Is it still provocative overweight anime horses and sexy lingerie model photography?" Usually it is...on the front page, at least. Having helped build this site into something-more-than-nothing I know that the front page, for all its purported attempts at sophisticated algorithmic-based sorting, is not really a reflection of the talents of the millions of members.
However, it is a reflection of what's hot right now. And what's hot right now is AI-generated imagery. DALL-E 2, Midjourney, and probably 2-3 others that escape me right now.
There are a number of reasons why you folks should not upload this stuff:
Under the ToS of most of these systems, you don't personally own the art. The art is owned by the creators of the software that generated it.
The art is generated by a system that scrapes the web looking for imagery it can categorize as something. It might find a painting of a figure right here on Deviantart, categorize that painting and its subject matter, and then used it to train itself. The image output you got was generated by a system lifting the work of your neighbors and acting like it did the heavy lifting. It's a collage-maker without the benefit of the collage being about something (like collage-makers try to say).
You did little more than type out a description of what you wanted. At best, you are a client who took the output they paid for, uploaded it to an art site and said "I made this."
The sheer popularity and ease-of-access of services like Midjourney means using them is going to be a populist activity—one that is inevitably going to delegitimize the work of others and whatever modicum of artistic integrity this website has. You see, when the front page is 50-80% AI-generated-imagery, a visitor is going to perceive the front page to be a representation of the majority of the content on the site. In other words, they see the front page as AI-generated and assume the rest of the library of work must also be this. This is human nature. Granted, before the AI-generated imagery problem, Deviantart's perception (again, based on the front page) was that it was all provocative overweight anime horses and sexy lingerie model photography. But at least that stuff was human-created.
Playing devil's advocate, I can just hear some comments and retorts. Allow me to get ahead of them:
What's the difference between an AI generator and a photographer? Certainly both rely on a machine to do the heavy lifting. Firstly, how dare you make this comparison. Secondly, an AI generator is a computer doing ALL of the lifting, while a photographer is using the camera as a tool. You'll notice that photographers have unique styles from one another, borne from years of experience, informed by their knowledge of lighting, the way skin and light complement one another, the human form, how to capture emotion, the power of color, figure-ground relationships, framing...the list goes on. Optimistically, AI generators are they themselves more akin to the photographer. By using MidJourney, you've hired a photographer to give you an image in their style.
Coming up with a clever description for an image output is itself the creative activity, isn't it? I will be the first to admit that creative writing is an art form. But what you are doing to build a MidJourney image is akin to typing a description for an ALT tag of an image for accessibility compliance. Creative writing it is not. Don't fool yourself.
Isn't this the direction illustration is heading in? I've seen this notion shared on Twitter—that publications, journals, books, news articles, and more will circumvent the use of human illustrators in favor of an AI generator. After all, why pay a person to make an image when a generator can pop one out for little-to-no cost? This might be true, for now, because it's novel. But an image generator will always be limited by the ability of the client to describe what they want to see. Like all clients, your descriptions will be boring as fuck, and fail to describe the impossible nuance and minutiae that come from only a talented human being putting "paint to canvas," so to speak. Also, recall how everything image generators pump out looks the same at its core (I mentioned that previously).
Isn't there something to be said about people finding ways to be creative when they otherwise can't draw, paint, photograph, write, etc? Yes I think it's wonderful people get excited by their artistic opportunities and explore those, but please refer back to the title of this overdue post: the AI art is not your art. You are the client.
K that's it. I wonder how this will age over the next 8 years.
Hey look, I'm tweeting!