Because I teach courses on digital society and digital storytelling at Milwaukee School of Engineering (MSOE), I have been reading and learning a lot about DALL•E, the new AI-based art generator, this fall.
I had seen references to DALL•E all summer on Twitter, where I follow several artists and graphic designers. But my interest kicked into high gear several weeks ago when I saw this article in The New York Times.
Of all the “complaints” I’ve heard about DALL•E so far (AI-generated “art” is not ART, etc.), the one that se
Since one of the best ways to learn stuff is to plunge right in, I decided to take the text of my “Haiku for October” post and pop it into the search/text “description” bar in DALL•E, just to see what happened. Below is my string of text and the resulting images.
I kind of liked these, but somehow they seemed a little bit dark. I especially liked the little glistening droplets in image number two, but I liked the colors, particularly the hint of aqua, in the last one. So I refined my text, and got these images.
These were very bright and cheerful. They are also a bit too abstract and maybe sport a little too much aqua and blue this time. To adjust, I removed the word “blue” and added “brown” and “realistic “ to the mix.
Hmm. These feel a little too green yellow and brown, not quite enough of the peach that I liked. But I do like these. A lot. In fact, I like almost every image that the platform generated this afternoon.
Of all the “complaints” I’ve heard about DALL•E so far (AI-generated “art” is not ART, etc.), the one that seems most valid and troubling to me is that DALL•E apparently has uploaded many artists’ work so that a “creator” can generate images “in the style of . . . ,” the concern being that these artists’ ability to own their style is being violated.
So I decided to further refine my text descriptions to produce images in the style of a few artists I like.
Here is my haiku, with my requested colors, in the style of Vincent van Gogh.
Here, in the style of Wassily Kandinsky.
Next, in the style of Salvador Dali. I could obviously go on and on, but as I assume the site’s name is a mash-up of the artist Dali and the movie/animated character WALL•E, he (Dali) seemed like a logical artist to try out.
As you can see, today’s artists probably don’t have all that much to worry about. Although I can sort of see these various artists’ “influence,” I would never look at one of these images and in any way think I was seeing a work by any of them. I’m not even sure that their “styles” manifest all that clearly, except maybe in the Vincent van Gogh, and there only because I already knew what I was looking at.
My main takeaway from this little exercise? DALL•E does seem like an amazing tool, and I think I have just found myself a new hobby!😄
Do you have any favorites among the AI-generated images above? Had you heard of DALL•E before? What do you think about the idea that anyone, including people with no art training, can now generate art with carefully chosen words?