DALL-E is the new artificial intelligence project from OpenAI that is sweeping the internet. It is an AI that can instantly produce a unique image based simply on a text description. There seem to be few limits, as the AI can create multiple high-quality images of just about anything you can think of. This has many people fearing that DALL-E will spell the end of human artists. But are the images DALL-E produces even art? Can AI ever create art?
First of all, what is art? It’s an age-old question, but I’ll do my best to answer it. To me, art is an intentionally created expression of consciousness. The intention of an artist is essential. A sunset in itself cannot be art, but someone who takes a photograph of that same sunset can be. It is the artist’s conscious perspective that turns something into art. In this respect, DALL-E’s images are not art, and AI can never make true art (unless AI becomes conscious).
So far as we know, AI is not conscious, and we’re likely far away from that being possible—if it is ever possible. The images DALL-E produces are not art in themselves because there is no conscious intention. But a human artist can use DALL-E to create art, just as humans have always used the latest technology available to create art.
Photography is a good example. There was similar hysteria in the past when the camera was first invented, many believing it would spell the death of visual art and put painters out of business. They were correct to some extent, as the camera did indeed put many painters out of business—but only the types who painted photo-realistically. When a camera could do the same thing faster, cheaper, and better, why pay a human to do the same thing? As a result, many painters shifted into more surreal and impressionistic painting—things a camera could not do—which resulted in some of history’s greatest artworks. Technology replacing human artists forces human artists to become more creative.
The second thing to consider with photography is that you cannot simply take a photo of anything and call it art. Photography developed into an art form wherein the photographer chose to capture a specific moment in time from a specific perspective that matched the consciousness of their own perspective. That conscious intention on the photographer’s part resulted in art, not the camera itself.
So how can the technology of DALL-E be used by artists to create new forms of art? Will the best writers of prompts be considered artists? Perhaps. There is certainly human consciousness and creativity that goes into writing the descriptions, and those writers who come up with truly novel conceptions and combinations of images might be considered art. But I don’t think purely AI-generated content will ever be art in itself. Instead, artists can use AI like DALL-E to create art of their own.
One example, that I would like to try myself, is using DALL-E to create a graphic novel. I’ve always wanted to adapt one of my stories into a comic book, but I am not a skilled enough drawer or painter. Even if I was, it would take an enormous amount of time to draw the hundreds of images needed for a graphic novel. That is time I would rather spend writing another story, as writing is my true artistic talent.
But if I could use DALL-E to instantly generate hundreds of different images from my story, I could easily turn a screenplay into a graphic novel. Each panel created by DALL-E would not be a work of art in itself, but the graphic novel as a whole would be the artwork. Naturally a graphic novel illustrated by DALL-E wouldn’t be as good as the best graphic novels, whose artwork is truly art in itself. DALL-E graphic novels would be more akin to concept art. Creating quick concept art (for movies, ads, companies, brands, etc.) might be the best use for DALL-E, to provide some ideas for the client before then hiring a human artist to create the final product.
Another use of DALL-E I would like to try as an author is to create book covers, and perhaps add some illustrations of certain scenes and characters throughout the story. In that case, like the comic book, the true art would be in the words, not the images. Writers have GPT-3 to worry about—though as a fiction writer I’m not too worried about being replaced by AI anytime soon. As I’ve said before, AI could potentially write formulaic genre stories quite well, but those are not the types of stories I am interested in writing.
Perhaps the best use of DALL-E would be to create images for blog posts, such as this one. Most bloggers (including this one) cannot afford to hire artists to create unique images for every blog post. Instead, they use free public domain images. Since most bloggers are already using the same stock images, DALL-E would likely present a stark upgrade to the status quo in online publishing.
It is unfortunate that DALL-E could end up replacing many human artists who would have created the images for book covers, blog posts, and other illustration and graphic design jobs. But like with painters before them adapting to photography, DALL-E presents artists with an opportunity to upgrade their creativity and produce images that AI cannot. In other words, create true art.
I wonder if DALL-E images will end up having an aesthetic of their own, so that people will instantly be able to tell it’s a DALL-E. For example, if you enter “tiger” into DALL-E, it will generate ten different types of tigers. But if 1,000,000 people enter “tiger” into DALL-E will it generate 10,000,000 different tigers or just the same 10 different tigers? (Or at least similar variations of those ten types of tigers.)
What worries me most about AI-generated art is the type of dystopian future Erik Hoel presents in his great essay about DALL-E, “AI Art Isn’t Art.” Hoel fears humans will still be making art in the future—much better art than AI—but AI-generated “counterfeit art” will become much more prolific in society because it will be cheaper and easier to create. As a result, most of human society will become so accustomed to seeing nothing but AI images that they will lose their taste in true art.
Some would say the deterioration of artistic taste is a trend that’s already been happening with each new form of technology—compare modern pop to classic rock to Mozart. Likewise, YouTube degenerated people’s taste for television, which degenerated their taste for movies, which degenerated their tastes for stage plays, etc. In a future full of AI-generated “art,” people may have to go out of their way to find true art consciously made by humans. But again, I would argue that is already the case today, as the most popular music, books, and movies are rarely the best or most creative.