This whole thing about AI art

What makes a person an artist, anyway?

What in the world is going on?!?

I’ve been oblivious to all the hype about AI (artificial intelligence) for a while now, but then got curious about it. I heard about Midjourney, a program that generates AI images with text prompts. So I thought I’d try it out.

First, a few disclaimers (or, TL;DR)

This is a very weird time in history, and AI is a hotbutton issue. I’m not sure how things will go down in the future, so maybe this post will “age like milk” eventually. But as of this writing, here is a summary of my views.

1. I stand in support with my digital artist brethren. As an oil and acrylic painter, I’m not personally affected by AI too much at the moment. But this whole thing concerns me very much.
2. I think there’s a great distinction between having the knowledge and ability to create the majority of the art yourself, vs. having something else generate a picture on your behalf based on an idea you have. I realize there are nuances here and some people devote a lot of time tweaking image details in AI.
3. AI image generation can be a lot of fun, as I learned when I experimented with MidJourney. I can see its appeal to non-artists as a sort of “toy” or “game.”
4. AI pictures are already being used in unethical ways, like scammers claiming that their AI images were originally genuine oil paintings and selling prints of these not-oil-paintings.
5. It’s too soon to know where things are going to go. Maybe AI will primarily be used as an additional “tool” for artists. Maybe it’ll destroy many jobs in the art field. I pray it’s the former rather than the latter.
6. AI is getting better all the time. The problems it has with anatomy probably won’t be as bad in the future. We shall see.
7. I’m not going to get too much into the controversy surrounding using images to train AI (against the original artists’ wishes) other than to repeat that I support my digital artist brethren who are most affected by AI.

AI and what does it mean to be an artist?

I’ll be honest with you. It’s really hard to reconcile the idea that someone can type in some prompts and out comes an image and this means they’re an artist.

Am I saying that they “shouldn’t” use AI for fun? No, I can see its appeal. It can be low-effort, instant gratification, what’s not to love?

The AI users who work long hours and paint over, modify by hand, and otherwise fine-tune their images with painstaking detail are one thing. There’s a lot of deliberate effort going on there, as well as a lot of decision-making about many of the details. This seems, to me, to probably be some form of artistry.

The typing of a string of words, even dozens or more words, isn’t on that same level, though admittedly, I’m not aware of the whole process. Even a few hours of fiddling and picking-and-choosing cannot possibly compare to the knowledge, training, and decision-making artists have historically had to make.

If someone wants to claim the work is completely theirs, I think they need to make all or at least the vast majority of the decisions during the creation of the image. If they let AI do a lot of the work, how is the result all “theirs”?

The above images are a Midjourney-generated portrait of a cowboy. The image on the right is the original, straight from Midjourney. The image on the left is edited by me, because I thought a few things didn’t “look right.”

You can see a few changes like the highlighted side of his nose, a shadow on his neck, the shadow under the brim of his hat, and a few other tweaks. There were other details about the AI image that I questioned but didn’t try to correct. (Who has all day?)

I didn’t understand why there was this deepening shadow over one eyebrow on the forehead. What was causing that shadow? When I got rid of it, the image made more sense. Also, the shadow that his collar cast on his neck seemed odd as well. When I smoothed that out, it seemed more natural. There were also some complicated shadows and highlights on the side of his nose facing the light. I didn’t understand why they would be like that, so I tried to make the lighting more consistent.

I’m not confident I made all the right calls in my attempt to correct this picture. All I know is that some things looked very off in the original picture. Perhaps more experienced artists won’t agree with the changes I made. (Entirely possible!)

What was the point of doing this? Because I wonder if someone with a more untrained eye (for example, someone with no prior training in drawing and painting) might let a lot of these weird little things pass. Relying on AI to get all the details right is a risk, especially if the person behind the prompts doesn’t have a lot (or any) prior art experience.

Even the most recent version of Midjourney creates images with details that look way “off.” This may remain an issue for quite a while; we shall see.

(Do you notice how all of the people are white, even though I didn’t specify in the text prompts that they should be white? Isn’t that interesting?)

(I sometimes didn’t list all of the text prompts used because I couldn’t be bothered to double-check exactly what I put in. Because who cares? This is totally low-effort.)

I think I’ll wrap this up. I still feel a little … ambivalent about this whole AI thing. I’ll be honest, I feel indignant at some of the attitudes expressed online. Attitudes that I consider to be disrespectful towards artists. However, I also feel like I need to wait a little longer to see how things develop.

Back to my art studies!



• Post