On AI Art
- 2021 Dec. 5, How Do We Value Art? What AI art means for tarot and oracle deck publishing
- Playing around with wombo.art
- Is AI art indistinguishable from human-made digital art?
- What principles determine great art?
- Socratic & Platonic philosophy on art
- Taoist & Confucian philosophy on art
- Art composition: design theory
- Disegno (Italian Renaissance)
- Confucianist aesthetics
- AI’s creative process vs. human artist’s creative process
- Is AI doing a better job than humans?
- 2021 Dec. 6, I Ching Oracle Cards with AI Generated Art (Free Download)
- Having fun with AI generated art apps
- AI art doesn’t “lack soul”
- Who is the creator when it’s AI art?
- Free download of I Ching oracle cards (AI generated art)
- Zip files of all downloadable image files for hexagrams
- 2022 Sep. 14, AI Generated Art + Tarot and Oracle Decks with AI
- Personal dabblings with AI art
- Does AI art lack soul?
- Rising popularity of AI generated art decks
- AI art won’t replace fine art (but may impact illustration)
- What artists are saying
- IP implications of AI art
- Valuing AI generated art
- More of my thoughts
Back in December 2021 I covered the topic of AI generated art and what it might mean for the marketplace of tarot and oracle decks here (“How Do We Value Art? What AI art means for tarot and oracle deck publishing“) and here (“I Ching Oracle Cards with AI Generated Art“).
But since then there have been new developments in this subject area so I thought I might revisit the topic.
Some Personal Dabblings with AI Art
Above to the left is a sketch I did by hand, first in pencil, then outlined in ink. I started with the following prompt, text I typed out myself and stared at for a good five minutes before putting pencil to paper: Solitude. Contemplating. Maiden in a moment of self-questioning.
I copied some text written by Hildegard of Binden on the transcendental experience of God, to fill the blank space. What you see took me two hours. Uh, tbh, probably longer than two hours. I lose track of time when I’m doodling. (The barely-there blue grid lines was added digitally, because that’s just something I like to do when I share my doodles to the public.)
What you see to the above right was produced via NightCafe, an AI art generator, with the same exact text as the prompt: Solitude. Contemplating. Maiden in a moment of self-questioning. I selected the art style “Charcoal” to see how close to a pen and ink sketch it could go. The illustration to the right took the program two minutes.
I’m fascinated by how similar the interpretations were, between me, a human, and AI tapping in to collective knowledge. In fact, in the past I’ve drawn illustrations in charcoal very similar to what the AI produced!
The pose, the facial expression, the way the hair falls, the vulnerability– if I rummage through my old art portfolio from high school, I can excavate a charcoal or pastel drawing that looks more or less the same with that!
Does AI Art Lack Soul?
I explored the question “does AI art lack soul” here in an earlier rumination on the subject. In that blog post, I talked about how this advent of AI generated art has shifted my former paradigm on the mind-soul relation.
This declaration you’ll hear oft repeated — AI art lacks soul; AI lacks soul — is one I’m most apprehensive about. Perhaps we can say we don’t understand the soul of AI, but to declare that AI art lacks soul… I dunno. It doesn’t sit right with me.
I’m not convinced that these works “lack soul.” If I’m getting all psychic and woo, I might say the impression of the soul that’s present feels different from a human sapient soul, just like an animal’s sentient soul or a tree’s soul feels different. You hear people critique the evident style or aesthetic consistent in AI generated art, but just because you don’t love an artist’s style or technical approach doesn’t mean that artist suddenly lacks soul.
So while I have many conflicting thoughts about AI art, the accusation that it lacks soul isn’t one of them. If anything, I wonder if the full body of AI generated art is mirroring back something deep within us collectively, for us to see.
A Rising Popularity of AI Generated Art Decks
Oh, and to illustrate what the community has been buzzing about with regard to AI-generated tarot decks (or in collaboration with AI) coming on to the market, I’ll feature several throughout this commentary.
When I first wrote up those blog posts, we were speculating on all the tarot and oracle decks illustrated with AI generated art that would be coming to market. Now in September 2022, we are officially seeing AI generated tarot and oracle decks on Kickstarter for fundraising. Well-respected tarot authors are trying their hands with these apps to produce tarot card illustrations.
I’ve been following the robust and opinion-charged discussions with great interest. I thought long and hard on whether I wanted to share my opinions. And then I also had to think long and hard on what my opinions even were. Because my feelings are so mixed.
On one hand, this is just the next phase of evolution in the world of illustration (and I say “illustration,” not necessarily “art,” because I don’t see this having any significant impact on the fine arts).
For funsies, below to the left is the Strength card illustration I drew for my in-progress Etteilla tarot deck. I then described my illustration as best as I could in a text input, which generated the image you see to its right.
AI Art Won’t Replace Fine Art (But May Impact Illustration)
Fine art in traditional media is still here, and still revered, but illustration by hand has in many ways gone obsolete. Right now, how many commercial illustrators are doing their work by hand? Most if not all of them (the ones who are productive at least) have gone digital.
Capitalism caused analog illustration to go obsolete, because if you’re a contracted illustrator and you’re still trying to do all your commissioned work by hand to paper, there is no way you can compete with the digital illustrators who finish their commissioned assignments in a fraction of the time. All the software hacks, shortcut functions, photo-bashing, tracing, and use of prior art underpaintings give digital artists a leg up.
There is also no denying that it’s easier to hide your flaws and technical weaknesses as a digital illustrator than via free-form analog drawing.
The vast majority of art you see on Instagram today is digital art. When I browse selections of tarot decks for sale, the majority are digital illustrations, not rendered in traditional media. And not all artists are transparent with how they are rendering their illustrations, so consumers are getting confused.
Digital art as a medium has not supplanted traditional media in the fine arts, but digital has unequivocally superseded and replaced hand-drawn illustrations.
Just look at tarot and oracle decks. I counted out the last 20 tarot and oracle decks sent to me by publishers, the latest ones on the market, and out of 20, only 3 were done in traditional forms of analog media (watercolor, pen and ink only, oil or acrylics, etc.).
The other 17 of the 20 most recently published decks were all rendered via digital collage or digital illustration.
Will we see the same trend happen with AI generated art? Oh, I’m sure of it.
Not even 30 years ago tarotists were crying “never” to digital art decks and now digital art decks have taken over the market. Or at the very least they’re trending like hot cakes at the moment of this writing.
A non-artist tarot deck creator will typically need to collaborate with an artist (a human being) to turn a personal vision into a deck of cards. This isn’t always easy. You have to pay the artist. Maybe the artist isn’t as cooperative as you had hoped. The artist’s progress is too slow. You don’t like some of the artist’s interpretations of your notes. Artists quit.
Now that we’ve got AI art generators, that non-artist deck creator no longer needs to collaborate with a human being. You can produce a tarot deck as close to your vision as you can get it on your own, at a fraction of the cost, and a fraction of the time. Not to mention you have more control over the process.
We level the playing field for all who want to transform their visions and interpretations of the tarot into published decks.
There is still a notable amount of skill required to produce a cohesive and beautiful tarot deck, even with an AI art generator. The text input instructions matter a great deal, and at least at this point in time, the best AI generated art still requires a few final personalized touch-ups by a human.
As for artists who have dedicated the last few decades of their lives honing their craft, how do they feel about AI generated art?
What Artists Are Saying
Based off my anecdotal observations, following online discussion threads on the topic and seeing who says what, generally speaking artists are not loving this new development.
Some art communities on social media have banned posting of AI generated art. If not outright banning, other communities are segregating AI generated works or enforcing strict policies of transparent disclosure.
I’m not against enforcing a policy of honest disclosure. In fact, I wish it was more customary in the tarot world. It’s considered best practices among artists to always disclose your media. Is it oil on canvas, watercolor, pen and ink, pastels, and what size is the canvas.
For reasons unknown to me, that best practice never transferred over to the world of deck illustration.
Sometimes I feel like I have to get all private-investigator-y, hunting down author background information, available art portfolios, and scrutinize the artwork to discern media– was this done by hand or is this digital art? And if digital art, just how much manipulation went into it?
If you are a deck creator using AI generated illustrations, would you call yourself an artist, or would you call yourself an art director?
Even when I drew the entirety of my First Edition SKT deck by hand and published it, I hesitated to call myself an artist, and went with the term “illustrator.” I hold the title of “artist” as sacred.
Since there is no official regulation of the term, I feel I need to self-regulate. If I truly do honor artists, then before I throw that term around willy-nilly, I have to think seriously about what qualifies one to self-identify with that mantle.
Now with the advent of AI generated art, where there is a clear imbalance between how much can actually be attributed to the human operating the program and how much can be attributed to the algorithm, is someone prolific at writing text instructions for the algorithm an “artist”? Are they even an illustrator? (Maybe.)
Intellectual Property Implications of AI Art
Finally, the intellectual property (IP) law around AI generated art is murky, confusing, and still in motion. Is it work-for-hire? I mean, who did you “hire”? Who should be credited with having done the work– the person who inputs keywords or the person who created the app? And if it’s a joint work, how exactly will the bundle of rights be divvied? (In recent times plaintiffs have even challenged the legal status of AI’s personhood, arguing that innovation invented by an AI should be patentable by the AI itself, and the copyright to AI-generated art should be owned by the AI.)
Is AI-generated art even copyrightable? Sure, we’ve got longstanding steps of legal reasoning, factors, and elements we apply for consideration of valid copyrights, but some of those steps of legal reasoning become outdated when faced with the specific conditions of AI and the metaverse.
How much of it is ownership under the app maker who can validly license the work to you, whereby the programmer argues that the IP to the app is theirs, and everything generated by their app is a derivative work they own? Or if the app is a mere tool or instrument, are you the bona fide copyright holder because you created it in as much as the IP rights conferred to an artist creating a digital collage? Is the work an “original” work or a “derivative” work?
How much of it is potentially copyright infringement based on the algorithms of some of these apps? How much of it constitutes web scraping? Could there be potential claims of unfair competition and unlawful trade practices?
When you use a graphics editor software program right now utilizing the program’s pre-set brushes and complex functions to create digital art, we don’t have any sense that the owners of the software own your art. You’re just using the program as a tool. What about an AI art generator program? To what extent is it a tool and to what extent is it creating the art and therefore the owner of the software program has credible arguments to claim IP ownership?
There are now programs where you can upload an artist’s portfolio of work to teach that program how to render art in that person’s style. Then you input text instructions for the program to generate art in that person’s style. What are the copyright implications and codes of ethical trade practices around that?
True, style or mere aesthetic cannot be copyright protected, but to what extent is it arguably trade dress and potentially protected by trademark? How are we going to differentiate, from a legal perspective, “General AI” models where the artificial intelligence is capable of self-learning, coming eerily close to acting on its own the way a human artist would, taking inspiration from preexisting work but effectively transcending it to create original art and Generative Art Systems, or algorithmic art?
What really is the difference between AI generated art via a model that self-learns and uses a unique single-case algorithmic process for each work vs. a human being creating digital collage art?
You mean to say the human being creating digital collage art is superior because human? Do we really want to go there as a society? Categorizing as superior vs. inferior based on immutable characteristics historically gets us into loads of trouble.
What is the authoritative law governing the IP of natural language processing (NLP), such as text-to-image machine learning models? Since there’s a big difference between intentional infringement and innocent infringement, how do we assign legal liability? And now that we’re global, what if the program developers live in what part of the world governed by one set of IP laws and the user lives in a different part of the world, governed by a different set of IP laws?
The current opinion, as of this writing, is that AI generated art is not copyrightable. This opinion is on the grounds that AI generated art does not meet the originality requirement for a valid copyright. It’s not “an independent creation” with a “modest quantum of creativity.” Another basis for this opinion is that the art you’re creating with these generators is a derivative work, based on preexisting original copyright-protected work, and therefore as a derivative, it doesn’t belong to you.
I’m not entirely convinced the legal opinion will stay that way for long. As an artist, someone who has been tinkering with a variety of AI art generators, trying to learn more about how they work, and listening to those in tech explain art generators, I do think in specific cases, there’s a compelling argument in favor of copyrightability for AI generated art. Esp. when the individual has added a quantum of creativity to meet the originality requirement, and there’s no scraping from preexisting copyrighted work.
The question boils down to whether the AI art generator itself is comparable to a pre-set digital paintbrush function on my software program that I use as a digital artist to create my work or is it removing the originality element necessary for copyright protection? I don’t think any of us are in a position to say for certain right now.
Valuing AI Generated Art
If an AI-generated illustration is technically better than a human-sketched one (in terms of proportion, composition, skill level of the detailing, etc.) and the AI version took 10 minutes while the human took 10 hours, what do you value more–the technically superior illustration or the one that dignifies human labor and original creativity?
More of My Thoughts
So what is my opinion? Have I even shared that yet?
There’s really no point in being in favor of or against AI generated art. At this rate, it’s an inevitability. It will displace human illustrators (but not the fine arts) and we are going to see more tarot and oracle deck illustrations, and book cover designs, and every aspect of commercial art taken over by AI, and the human element will be the operator of that AI program.
When a tarot or oracle deck created from AI generated art comes to market and I’m to review it, I’ll review it as I would any other deck. I can only hope that creators will be honest and disclose how their illustrations were rendered. There should be an accompanying artist statement that shares the creative process.
To which I then have to interrogate myself, why does that matter? My answer would be that it matters because it tells us the extent of human labor, study and self-cultivation, and original creativity it required to render that art.
And then I realize–well, there you have it. That’s my opinion. I care about the creative process. Whether I’m looking at an oil painting, digital painting, or AI generated art, what was the creative process that went into it? Then we evaluate that creative process alongside the resulting work.
So what about you? Have you been following the evolving discourse on AI art?
12 thoughts on “AI Generated Art + Tarot and Oracle Decks with AI”
As always, your insightful explorations of these topics prompts me to a do my own thinking as well. AI is definitely a topic of conversation in the artist community, but so far, most of my friends and colleagues are still at the stage of playing with this new medium.
As an artist, I do think of AI as a medium, like any other. Honestly, I think the controversy around it is not different from the controversy when photography became popular. Photography did fundamentally change art, and there was a lot of talk about whether painting was dead, replaced by photography, and even whether artists were suddenly redundant, since anyone could take a photo of something.
But we still have artists and paintings and drawings. We even still have realist painting. After the initial panic of change, once the new medium was absorbed into the arts, it found its place among all the other media. I think this will happen with AI, too. We’re nowhere near that point yet, though.
One art form really does not replace another, because all are tools of the artist who will use whatever they need, however they need, to express the messages within them.
I think capitalism is a much bigger issue in the arts. You mentioned how professional illustrators really can no longer work in analog media because of the fast turn-around demands of the publishing industry. That is something completely unrelated to art, i.e. corporate profits, interfering with the arts. It doesn’t so much promote digital art as a medium as it restricts what kind of art the public gets to see in illustrated works. I think of it as censorship by deadline. Money censorship, rather than the political kind. It binds the art to the service of a profit margin rather than the stories being told. The capitalist system can have the same toxic effect on AI art, perhaps even more obviously, if it takes over AI art production to mass-produce art-like products for cheap sale – or worse yet, to lend legitimacy to the NFT market.
Artists who want to make good art with AI should work and experiment with it the same way we do with analog media, without regard to what is “popular,” since, more and more, it’s hard to tell what is really popular and what is just being spoon-fed to the public by corporate gatekeepers.
LikeLiked by 4 people
Definitely interesting! With regard to AI being conceptualized as just a medium, and therefore something created with an AI art generator by a human artist should be considered authored and owned by that artist– I wonder about that. I don’t think I have any conclusions. For example, Jasc Paint Shop Pro has this function where you can auto-transform any image into stunning abstract fractal art. But if I use that function to produce a work of fractal art, I certainly would not try to pass it off as my own artwork. I wouldn’t say, “Hey, I did this!” And I kind of see AI art generators as being along those lines, rather than a medium like acrylics, or even photography.
Anyway, yes. So much to think about and not a whole lot of answers yet! =D
LikeLiked by 2 people
This is going to be a huge deal. In comics, there has been a huge discussion over this. This is going to impact comics artists tremendously and they’ve gotten the short end of the stick for years- especially when they’re not getting credit where credit is due for the characters they’ve created. Now to have this. I think AI art can be useful if someone can’t draw, like me, but doesn’t want to engage an artist for some of the artwork. However, it means being upfront honest on how the artwork is made and what kind of rendering goes into it.
I’ve got skills in Photoshop, Illustrator, and InDraw. I letter comics, colour digitally, and create graphics for people. It’s part of what I do outside of my day job. So, there is a lot I will STILL do on the creative side of things- especially since I tend to be hands on. But my drawing skill is stick figures, maybe some dabbling in faces. My goal has always been- pay an artist for things I can’t do, make what I can, and go forward.
AI art can help, but it’s not the end all be all. Why? So much of it seems similar in many regards. I love artwork that hits different. Whether by colours, style, or a combination of things. AI art though it can be create or mimic others- it can only put out what you put in and it can find. So, for me- it’s more for basic pieces I’d want for thematic resonance. Or when I do my basic ideas to see how I want to work them before getting the artist so they can see my idea- because really- I suck at drawing.
I find AI art worrisome in many ways. For myself, I limit how I would use it for me and why I would use it, if I choose to do so. For others, I would hope they’d see the same core issues with AI art, realising the inherent problems, the need to be fresh and new within themselves, and stretch beyond AI art for the things which are truly important to them. I am pro-artist and always will be. I have so many artists I support and I can’t see that changing.
LikeLiked by 1 person
I agree with the proponents who say that much of the output quality (and uniqueness) rests in how skillfully the text instructions are crafted. A lot of us are generating very “same-same” style AI art because we’re still limited in our skills with how to input the text. I’ve seen some amazing AI generated art that would be hard to discern from human created digital art. So this tool is certainly going to be a big skill boost for many artists and illustrators. It’s interesting to wonder, though, is it a cheat?
LikeLiked by 1 person
I think a lot of the “same-same” quality of AI art is due to it not actually being AI, Artificial Intelligence. These art generators are just fancier versions of the text generators that have been around for years. They pull image elements from built-in databases, and they are limited by how big and varied their database is. Even the Jason Allen image that won the award has the same-same color palette, line style, textures, etc., of pretty much all AI generated art. You mentioned programs to upload an artist’s work to the generator so it can learn that style. I wonder if it really does learn, or if it just breaks down the images to data points to add to its database. At the level of the technology now, it cannot produce original art, just randomized content.
You also brought up the ethics and legality issues around copyright, and I completely agree. Like the labor displacement issue of the commercial art industry, I think this is a very important point we should all keep in mind. Without regulation and transparency, these generators could be steered into pirating art, stealing especially from digital artists. There’s a lot of profit in digital piracy, as we all know too well.
It’s kind of an argument in favor of hand-made art. A colleague of mine once quipped that the only kind of art that is safe from copyright infringement is the kind that can’t be cheaply copied. He said, “Go ahead, forgers, replicate my anatomically perfect, life-sized, paper sculpture of a hummingbird skeleton. Have fun putting in the 600 hours it takes.”
I come from a surrealist background. Surrealism, dada and other modernist movements have always used found images, automatism and automation, and other methods of randomization. You mentioned in a comment that, if you used an AI generator to fractalize someone else’s painting, you would not claim that as your own art, for ethical reasons. That’s completely legitimate. I support that view.
However, to me, it also sounds a bit like that time Marcel Duchamp drew a mustache and beard on a postcard of the Mona Lisa and titled it “L.H.O.O.Q.” Like many such “artistic pranks” in surrealism, the art of L.H.O.O.Q. wasn’t in the object/image. It was in the artist’s message, the social commentary and implications, and the emotional response it triggered in viewers. But before people start claiming fair-use license to use other people’s images, it should be noted that the Mona Lisa was well within the public domain when Duchamp pulled his prank, in 1919.
Actually, the only point you’ve made that I don’t really agree with is this: “[W]hat do you value more–the technically superior illustration or the one that dignifies human labor and original creativity?”
I do think we need to ask ourselves how much we value art that dignifies human labor and creativity, but I think you’ve got a false dichotomy in the question. I do not believe AI offers an alternative of a “technically superior” illustration. How is the automated art superior, and to what? It’s only superior in terms of being cheaply mass-producible, but it’s hardly “technically” superior as art. It’s debatable how applicable that idea even is to art, generally. I think the real choice is, do we want cheap decoration, or do we want art?
For myself, I don’t have digital skills, so all my work is hand-made. I’d never be able to get work in today’s increasingly automated commercial art market, and that’s just as well, if your prediction is right, and commercial illustrators are doomed to become obsolete. I’m currently writing/publishing a web novel for which I’m doing my own illustrations, for online display and, someday, print. They’re all hand-made paintings on paper, uploaded by me, and they take time, but no one is counting my man-hours but me, and I think the labor is worth it. I do not think my images would translate will to any AI generator’s database, so it really will come down to whether the audience wants those slick, limited palette AI images, or they want my textured, irregular, handiwork.
LikeLiked by 1 person
I just laughed out loud at my desk, thinking it would be fun to pull an Andy Warhol on these AI images by creating them, printing them, gluing them to canvas, and painting over them by hand. Warhol would have painstakingly replicated them in paint as portraits of the objects (like his tomato soup can painting), but I’m lazier than him. I’d just draw over the “original.” Then I’d further imitate Warhol by mass-producing prints of my painting. I may actually do that. It would be funny. Let’s everybody do it.
LikeLiked by 1 person
They are exactly “art generators” rather than any kind of AI as gets popularly conceptualized, and therein lies the problem. Unless you’re an actual AI or machine learning specialist, it’s very easy to get fooled by what things-called-AI can do. Especially because the loudest and most famous people talking about AI actually know very little about it — Elon Musk, to name the first example that came to mind, has an academic background in physics and economics; whatever programming stuff he did as a childhood hobby or later as an intern was far removed from AI or machine learning. And yet his opinion and understanding of AI holds an outsized influence on the popular discourse, so in the end the misconception propagates faster than any genuine understanding, because at some point it just becomes part of the background “knowledge” we pick up via osmosis and never stop to critically examine.
LikeLiked by 1 person
I think ai still got a long way before the art world needs to worry, sure you can make a good looking cloud or background but it’s still looking wonky when trying humans or other living creatures doing anything but standing straight and alone, and most of us needs an artist in order to show relationships between people.
I found out about the Midjourney service a few days before reading this post, but I didn’t know about the other services you mentioned. I’ll have to check them out.
Thanks for showing us the art you’ve done with these services. The pictures all look great! I’m also doing I Ching cards to help me get a better feel for the hexagrams. My cards aren’t as nice as your, but I’ll blame it on just getting started with the service and no post work.
Here’s a sample, it’s #11, Tranquility.
Wow! This is super cool!
LikeLiked by 1 person
Pingback: How Do We Value Art? What AI art means for tarot and oracle deck publishing – benebell wen
Pingback: I Ching Oracle Cards with AI Generated Art (Free Download) – benebell wen