Ridiculous Opinions #234!
So, yesterday, Tracey left for India for Senior Fine Arts. Whenever Tracey leaves that allows me time to jump into some kind of obsessive activity that I would never do while she was here. Last year when she went to Canada, I wrote and recorded 12 songs (still sitting on my hard drive, thank you very much). One time when she left, I wrote a movie. Basically, it's an unhealthy way of accomplishing a lot in very little time, because otherwise, I would be sitting around, watching TV and that's no good for anyone.
So, I came home from work yesterday and I was thinking about how I would occupy my time. I'm not working on any big project at the moment, so there weren't a lot of options. But then, I checked my email and saw one from Google that talked about how Google Bard (Google's AI system) was able to make images from text. All you had to do was input a sentence and the AI would spit out an image for you.
This was intriguing to me, so I bopped on over to the site and gave it a shot.
I'm not a huge fan of AI and especially not a huge fan of AI art. Sure, it's amazing and all, but after you play with it for a bit, you come to realize that it is limited. For your average person, it's pretty cool to type in "Draw me an otter riding a skateboard" and have the computer spit out a photo-realistic image of that very thing. But after a little while, most people get bored with it and the novelty wears off. Why?
Because most people are simply not creative.
And it's difficult to be a creative person. It's work. It involves an inordinate amount of thinking to be a creative person, which is an anathema to your average person.
Most people are quite content to get on their phones or watch TV, because that's the opposite of thinking. It allows them to be passive; to shut their brain off for a while; to relax. Sometimes, this is necessary because of the stress of colleagues, a job, or life in general. It's not wrong to want to shut down for a while; it's just that if you want to be a creative human being, you can't do that all the time.
So, though AI art and writing are amazing, it doesn't take long to realize that after the novelty wears off, you still have to think! This, of course, was where I was at when I went to Google Bard yesterday to put it to work.
Now, I'm not one to go easy on these machines. I'm not going to type in "Draw Elmo surfing in Hawaii." That's too easy. So, I tried this prompt:
Create a comic-book drawing in the style of Arthur Adams of a group of apes (orangutans, baboons, gorillas, and chimpanzees) all sitting around a round table, playing cards. Each of them should be dressed in the same type of futuristic military uniform and large weapons should be resting casually around them. The biggest ape should be the gorilla, who is seated in the center of the picture. He is smoking a cigar and looking at the cards in his hands. The apes are illuminated by a single light above them, with a square-shaped, brown shade. They are in an apartment that is crumbling from years of neglect.
Not an easy prompt. And guess what? Google's newfangled image-maker gave me this in reply:
Surprise! I tried about ten different variations of this. I simplified, I expanded, I eliminated. Google just wouldn't do it.
But what about Microsoft's version? I thought.
Google uses its own, proprietary LLM. Microsoft licenses theirs from OpenAI, which created ChatGPT. Would it be able to do better? Here's what it spit back at me...
Holy crap! Pretty good. Not all of them were good, but this one was pretty spectacular.
I tried a couple of variations on this and each of them were right along the lines of what I wanted.
Then, I began to wonder...
A few years ago, I set out to draw one of the comic scripts that I wrote. It was a complex script about multiple realities all converging at the same time in New York City. I spent a lot of time drawing the first issue and I absolutely HATED it. Though I can draw, I cannot draw what I see inside my head. I know what I want, but I can't ever translate that to the page. What ends up happening is I spend a lot of time on my art and then I end up hating the results. This is typical of artists the world over and is part of our general self-loathing.
But as I was messing with Microsoft's program, I thought, What if I use AI to augment the art that I already have?
An intriguing concept.
What if I asked the AI to draw things that I had already drawn, but was unsatisfied with? What if I asked the AI to FIX the things in my drawings that were BAD?
Now, I have to be BRAVE to show you my crap artwork, but here you go...Here was the original first page, as done by me...
The little boy that I drew is god awful. And there are other problems. It's easy to see where I spent a great deal of time (the third panel is me at my best). The fourth and fifth panel is me just wanting to get something done. Because I worked on this page on different days, the quality varies. And, as you know, I have a day job. What you see is what I was able to do when I had the time.
So, I went about adding prompts. I didn't like the way I drew the boy, so I had the AI produce a new version of that character. It did. I was happy.
I took the drawing that the AI produced and I inserted it into my own artwork, making modifications as I went through to make it blend more easily.
The prompt was this:
Create a wide-screen, black and white comic book image in the style of arthur adams of a young boy, standing on the streets of Manhattan. The boy should be wearing a striped shirt and thick, black glasses that obscure his eyes. It should be a medium shot of the boy, with his father on his right and his mother on his left and both of them should be pointing away from him in different directions. A reflection of a light above him should be in the boy's glasses.
It spit out a few pieces of art that looked like this...
Not bad, but not what I wanted. The AI was unable to capture the style of the artist that I chose (Arthur Adams is incredible). The artwork that it consistently spit out was more along the lines of Charles Burns. There's a lot of cross-hatching and shading that makes it look like a woodcut or an illustration for a novel.
But I kept tweaking until I got what I wanted. I finally had something decent that I thought I could use. So, I inserted that artwork into my own and produced this:
And I was left with the confusing question:
IS THIS WRONG?
I don't have the skills to produce good artwork. I know this. I've known it for years. My art is a good parlor trick. It's enough to wow a few people, but not enough to be professional or to gain any kind of respect. I'm a writer. I would much rather have someone else produce the art for my books.
But I don't have the money to pay an artist. I don't have the time to acquire those skills. I have chosen a different path in life and at my age, the ability to gain the skills that I need to become a good artist is precluded by the fact that I have a day job that occupies more of my time than I'd like it to.
AI enables me to do the things that I'd like to do, but don't have TIME to do. That's still my story. Those are still my layouts. The concept of the little boy character is still exactly what I envisioned. Where does the line get crossed?
The first panel was drawn completely by ME. But I used photo reference. I traced it. That happens all the time in art. The third panel is an amalgam of three different photos that I used as reference. I didn't take those photos. I blended them together to produce something new.
The awesome cross-hatching that the AI did on its images? I found a brush in Procreate that will do exactly that in an instant. So, I can go through my art now and make it all look meticulously cross-hatched whereas in reality I faked it.
How is any of this different than using AI?
If you're looking for answers to these questions, I don't have them. I ended the day more confused than ever.
AI art is a problem. AI video is a problem. But is this simply another version of the printing press, where we are so used to one method of doing things that we can't accept a new way? And are we really going to be able to put this cat back into the bag? The jobs of many artists are going to be lost because of AI. But will they ever come back? The answer is: Probably not.
I worked until I was sick of looking at these images yesterday. I managed to improve several pages of my artwork with work that was infinitely better than what it was originally. Will this book ever see the light of day? Probably not.
I think AI is a big, self-perpetuating myth. It's not as good as everyone says it is. It's a magic trick. It's smoke and mirrors. It will only ever produce average writing. It will only ever produce lifeless artwork. To your average person, it will, perhaps, help them write a better cover letter or produce an essay that will get you a "B" in your English class.
But in the hands of the right people, could it be something more? Should artists consider this a new tool in their arsenal?
I leave you with this final quote from Banksy...
The greatest crimes in the world are not committed by people breaking the rules but by people following the rules. It’s people who follow orders that drop bombs and massacre villages.
Take from that what you will...
