So THIS is what has kept you out of trouble (?) since Wednesday. Fascinating! And your conclusions feel like a bit of a relief. Wanna discuss copyright?
A friend and I were trying to get less biased results from Midjourney and it really resists efforts to avoid specific, culturally-hot stylizations. "Woman with dark hair and eyes" produces an army of clone girls (clearly between 10 and 18 years of age) with the same shape mouth, face, nose, light dusting of freckles, and..... LIGHT EYES.... The darkest eyes are medium caramel hazel. Similar attempts to prompt a male face (no race specified) with dark eyes gives you unlimited portraits of Timothee Chalamet from different angles and differing amounts of tanner applied. No variants ever stray from this weird idealized facial standard. Apparently the training the AI gets is subject to an enormous amount of bias, whether conscious or not.
This would be due to the majority of images on the internet, but also due to tagging of eye colour not being common, similar to the #hands problem. Nevertheless, try “Woman, brown eyes.” The comma seems to help (?), and remember that it recognizes colours, whereas “dark” is probably something it associates with mood/overall colouring. Your bigger problem is with “woman” or “man”—that seems to be where AI is hopelessly polluted (with samey-looking people of a young age and type), likely due to gazillions of advertisements and also selfies with samey-looking poses/filters.
I know you probably already got there.... The "intelligence" being assembled seems to have its own very specific profile, complete with age, generation, media diet, race and class.
Marian, I am a HUGE fan. (I have all your books, too.) Glad to see you exploring the AIs. I think you'll find this piece I wrote for Wired about image generators to be resonate with your notes here. Particularly its kin to what we do when we photograph. https://www.wired.com/story/picture-limitless-creativity-ai-image-generators/
This is one of the most interesting things I've read about AI in a while. I've felt a lot of conflict about AI image generation since it came into the public consciousness this summer. I'm a game designer who relies on freely-available public domain art and stock images to illustrate my games. When I find something good, I feel a thrill similar to the one you discussed here.
When I first encountered Dall-E, I was like, "This is amazing! I can make awesome art for any project I want!", but it quickly left me feeling hollow. I'm not sure if this is because of the uncanniness of the images—there's always something slightly wrong about them—or because the ease of generating them robs me of the collector's feeling of discovery. After much back-and-forth with myself, I've decided that I'm okay using AI generated art as a substitute for clip art, but otherwise I'm going to stick to what I can find in the public domain. And I've also started learning to draw as my own little protest in favor of human-created art.
Thank you, and I LOVE that you’re learning to draw! When I first used Midjourney I had a strong urge to sit down and start painting, not exactly in protest, but … I don’t know why. I now think about painting all the time—sadly, finding time for it is much harder. But I’m wondering if playing with Midjourney is triggering something in the brain—a creative itch of some kind that the AI is not satisfying.
That's fascinating! I had the exact same thing about drawing—I'd never thought very hard about it and suddenly this fall it was on my mind all the time. Found a great class on skillshare and have been drawing every day. Lately I've been thinking a lot about how when you're making any type of art, you're constantly doing things and thinking, "Do I like this? Yes, great keep going." or "No, do something else." Those thousands of tiny refinements, trying to make something that pleases you, are similar to what AI is doing but radically different, because AI doesn't have taste that it's trying to satisfy. I think this is why AI art strikes me as hollow—it lacks the thing that even bad art has, which is the knowledge that someone made something this way on purpose because they liked it.
Saw this article in PRINT magazine and came here to leave you a comment.
It's great to see an artist's take on MidJourney, and to read your conclusions about the strengths and weaknesses.
I've been exploring and fighting with MidJourney and, like your friend, giving in or giving up pretty regularly. But it can conjure some terrific stuff.
The bias is very pronounced. One day I decided to create images of gray hair - long, beautiful gray hair. After a few iterations, I thought I should add some ethnic diversity but no matter what prompts I added (except for African American) I got the same young, pert Caucasian faces. Hmmm.
Another weirdness: I was creating images of vegetables and fruit and I added "glamorous lighting" because I wanted very attractive food. Instead, I got images of a young, pert Caucasian girl. Hmmm.
Last weirdness: I was creating Valentine's Day images - plump Valentine's Day hearts - and MidJourney also included an image of a Corvette with flames coming out of the sides. ??? Hmmm.
Keep reporting your explorations, findings and conclusions.
So THIS is what has kept you out of trouble (?) since Wednesday. Fascinating! And your conclusions feel like a bit of a relief. Wanna discuss copyright?
Yes it is, and Yes! Copyright is next, so if you want to talk, call me later!
Great piece... may I share this link and a few of your fantastical images?
Thank you and yes, Absolutely!
A friend and I were trying to get less biased results from Midjourney and it really resists efforts to avoid specific, culturally-hot stylizations. "Woman with dark hair and eyes" produces an army of clone girls (clearly between 10 and 18 years of age) with the same shape mouth, face, nose, light dusting of freckles, and..... LIGHT EYES.... The darkest eyes are medium caramel hazel. Similar attempts to prompt a male face (no race specified) with dark eyes gives you unlimited portraits of Timothee Chalamet from different angles and differing amounts of tanner applied. No variants ever stray from this weird idealized facial standard. Apparently the training the AI gets is subject to an enormous amount of bias, whether conscious or not.
This would be due to the majority of images on the internet, but also due to tagging of eye colour not being common, similar to the #hands problem. Nevertheless, try “Woman, brown eyes.” The comma seems to help (?), and remember that it recognizes colours, whereas “dark” is probably something it associates with mood/overall colouring. Your bigger problem is with “woman” or “man”—that seems to be where AI is hopelessly polluted (with samey-looking people of a young age and type), likely due to gazillions of advertisements and also selfies with samey-looking poses/filters.
I know you probably already got there.... The "intelligence" being assembled seems to have its own very specific profile, complete with age, generation, media diet, race and class.
Marian, I am a HUGE fan. (I have all your books, too.) Glad to see you exploring the AIs. I think you'll find this piece I wrote for Wired about image generators to be resonate with your notes here. Particularly its kin to what we do when we photograph. https://www.wired.com/story/picture-limitless-creativity-ai-image-generators/
Thank you! I will read your article soon—along with a bunch of others i have on the pile … thx!
Very interesting questioning read.looking forward to the next one.
I love the baby seal ❣️
This is one of the most interesting things I've read about AI in a while. I've felt a lot of conflict about AI image generation since it came into the public consciousness this summer. I'm a game designer who relies on freely-available public domain art and stock images to illustrate my games. When I find something good, I feel a thrill similar to the one you discussed here.
When I first encountered Dall-E, I was like, "This is amazing! I can make awesome art for any project I want!", but it quickly left me feeling hollow. I'm not sure if this is because of the uncanniness of the images—there's always something slightly wrong about them—or because the ease of generating them robs me of the collector's feeling of discovery. After much back-and-forth with myself, I've decided that I'm okay using AI generated art as a substitute for clip art, but otherwise I'm going to stick to what I can find in the public domain. And I've also started learning to draw as my own little protest in favor of human-created art.
Thank you, and I LOVE that you’re learning to draw! When I first used Midjourney I had a strong urge to sit down and start painting, not exactly in protest, but … I don’t know why. I now think about painting all the time—sadly, finding time for it is much harder. But I’m wondering if playing with Midjourney is triggering something in the brain—a creative itch of some kind that the AI is not satisfying.
That's fascinating! I had the exact same thing about drawing—I'd never thought very hard about it and suddenly this fall it was on my mind all the time. Found a great class on skillshare and have been drawing every day. Lately I've been thinking a lot about how when you're making any type of art, you're constantly doing things and thinking, "Do I like this? Yes, great keep going." or "No, do something else." Those thousands of tiny refinements, trying to make something that pleases you, are similar to what AI is doing but radically different, because AI doesn't have taste that it's trying to satisfy. I think this is why AI art strikes me as hollow—it lacks the thing that even bad art has, which is the knowledge that someone made something this way on purpose because they liked it.
Saw this article in PRINT magazine and came here to leave you a comment.
It's great to see an artist's take on MidJourney, and to read your conclusions about the strengths and weaknesses.
I've been exploring and fighting with MidJourney and, like your friend, giving in or giving up pretty regularly. But it can conjure some terrific stuff.
The bias is very pronounced. One day I decided to create images of gray hair - long, beautiful gray hair. After a few iterations, I thought I should add some ethnic diversity but no matter what prompts I added (except for African American) I got the same young, pert Caucasian faces. Hmmm.
Another weirdness: I was creating images of vegetables and fruit and I added "glamorous lighting" because I wanted very attractive food. Instead, I got images of a young, pert Caucasian girl. Hmmm.
Last weirdness: I was creating Valentine's Day images - plump Valentine's Day hearts - and MidJourney also included an image of a Corvette with flames coming out of the sides. ??? Hmmm.
Keep reporting your explorations, findings and conclusions.