AI Art Part 2: What kind of world do we want?
In the second part of this piece, we will consider the rise of AI-generated art from a more subjective point of view, focusing on its ethical and societal implications. In the first part found here, we went over why AI models do not store and reproduce exact copies of the artworks they have been trained on.
Thank you to 'Yote, who has a PhD in computational biology, for providing feedback and fact-checking for this article.
The backlash against AI art endangers the hope for a world where ideas are shared freely
Stable Diffusion contains unauthorized copies of millions—and possibly billions—of copyrighted images.
-Stable Diffusion Litigation
The second major charge levelled against AI is that it is being trained on copyrighted images, without permission, and that only public domain or explicitly-allowed images should be used as training data. I believe that this is both an unreasonable demand, which is not applied to humans, and harmful to the very artists it claims to protect.
Imagine one's eyes worked like a video camera, recording 60 frames per second, and one were awake for 18 hours a day. That would be almost four million frames per day. Images of people, buildings, cars, animals, trees, the sky, films, paintings and so on. All of them under different lighting conditions and seen from different angles. If one attends a furry convention where artists are drawing, one will often see that they have a second page full of reference pictures from different perspectives. Animators will watch video of whatever it is that they are trying to draw, perhaps they willeven bring in live animals to study how they move in person. These experiences are how we know what the world looks like and form the raw material that artists use to create their art.
But, does anyone believe that those reference images used by human artists are all free of copyright? How many artists search for reference photos of animals and then also check the copyright status of those photos, using only the copyright free ones as references? I would be surprised if even a single one did that. Do we believe that human artists can watch a copyrighted film and not have it influence their own work? I know, personally, that if I've spent time reading a good book, it tends to influence my way of writing and thinking immediately afterwards. Why are we trying to hold AI to a standard to which we do not hold human artists? To a standard that we would find completely ridiculous if applied to human artists?
ten years ago today Aaron Swartz was driven to suicide after being hounded by the state for making scientific knowledge accessible to all, and devoting his life to the cause of internet freedom from corporate dominationpic.twitter.com/W0T1gPwcE1
— ☀️👀 (@zei_squirrel) January 11, 2023
While I am generally sympathetic to concerns about AI being trained on copyrighted images and then used to generate commercial images, it is important to note the training is not done with the intention of copying any specific image. Instead, it is done with the intention of learning how to draw different concepts. One may copyright a specific drawing of a fox but one can not copyright the concept of a fox.
What would the furry fandom look like if we demanded that people could not learn from or copy the characters and imagery they see in copyrighted material? Are we saying that artists should not draw those images? Every furry artist that has drawn Pokémon, Digimon, My Little Pony, Zootopia, Helluva Boss, How To Train Your Dragon or Puss In Boots fan art has done so using copyrighted materials. Even some artists who have complained about AI using copyrighted works, have themselves drawn art from copyrighted works. I think that is hypocritical.
Fan content necessitates using copyrighted materials and building on them. This is a great way to learn and gain skills! Some people might think that it's not proper art but I would strongly disagree; the over-one-million words of Fallout Equestria: Murky Number Seven rival almost any "original" English literature. There's never been a strong division between fan creations and original works. Lewis Carroll's 1865 novel Alice's Adventures in Wonderland is great. So was the 1951 Disney animated film adaptation. And the 2000 American McGee's Alice video game. Peter Jackson was a fan of JRR Tolkien's Lord of the Rings long before he adapted it into an epic trilogy for New Line Cinemas. RJ Palmer got to work on Pokémon: Detective Pikachu because of his Pokemon fan art. Whatever criticisms one may have of its literary value, 50 Shades of Grey was originally Twilight fanfiction. We are meant to build on one another's ideas!
Copyright is important for creators but it needs to be balanced with how it impacts society. Modern day copyright protections have been significantly extended. Originally, in the US, copyright lasted for 14 years with the possibility to extend it by a further 14 years if the author was still alive. Presently, US copyright applies for the lifetime of the author plus an additional 70 years and, due to the influence of powerful companies, notably Disney, corporate copyright extends for 95 years!
I fear that the backlash against AI art threatens creativity, both in the furry fandom and broader society, by normalising even stricter interpretations of copyright. Already, copyright is used to suppress the creativity of fans and artists. In contrast to the RJ Palmer story, I know of several furry artists who received cease and desist letters from Nintendo which forced an end to some amazing comics. The same thing happened to a fan-made hack called Pokemon Prism in 2016. Similarly, a crowdfunded , fan-made Star Trek film was shut down by CBS in 2015. However, importantly, even using public domain materials can get your videos a copyright strike from overzealous algorithms and copyright trolls on Youtube. It has happened to me and it recently happened to popular violin duo TwoSetViolin. This is not copyright being used to increase creativity or to protect creativity; this is copyright weaponized by corporations. Artists lose out. Society loses out.
Not only corporations, but also individual artists can be overly-protective, not only of their own artworks, but even of their own art style. This is crazy to me. Artists have been adopting each others' styles for centuries; Impressionism is essentially a lot of people copying other artists' style. (The name "Impressionism" comes from Claude Monet's Impression, soleil levant, although he was not the sole originator of the style.) Characterising an AI trying to learn a specific art style as art theft is thus not justified, and does not make sense either: Theft requires one to deprived of their property; making a copy or using an artwork in a training set does not deprive the original artist of anything.
Rather than turning to the tools of large corporations to restrict and control, we should aspire to help one another. I see a better role model in the free software community which advocates the four freedoms:
• The freedom to run the program as you wish, for any purpose (freedom 0).
• The freedom to study how the program works, and change it so it does your computing as you wish (freedom 1). Access to the source code is a precondition for this.
• The freedom to redistribute copies so you can help others (freedom 2).
• The freedom to distribute copies of your modified versions to others (freedom 3). By doing this you can give the whole community a chance to benefit from your changes. Access to the source code is a precondition for this.
Many of those freedoms could be adapted for artwork and would allow the use of artwork to train AI models. None of us, artist or otherwise, lives in isolation. We are part of a community and, as members of the furry fandom, an even more close-knit community than many enjoy. We all learn from and are influenced by one another; our ideas were built on the ideas of others and, in turn, we should allow others to build on ours. Anarchist philosopher Peter Kropotkin made this point well.
Every machine has had the same history — a long record of sleepless nights and of poverty, of disillusions and of joys, of partial improvements discovered by several generations of nameless workers, who have added to the original invention these little nothings, without which the most fertile idea would remain fruitless. More than that: every new invention is a synthesis, the resultant of innumerable inventions which have preceded it in the vast field of mechanics and industry.
Science and industry, knowledge and application, discovery and practical realization leading to new discoveries, cunning of brain and of hand, toil of mind and muscle — all work together. Each discovery, each advance, each increase in the sum of human riches, owes its being to the physical and mental travail of the past and the present.
By what right then can any one whatever appropriate the least morsel of this immense whole and say — This is mine, not yours?
Artists today benefit from centuries of effort by artists who developed the techniques, colour theory and other tools which are still used today. Modern artists draw using computers and tablets which have been built and refined by countless minds. Digital artwork is drawn with software written by others and, when using programmes like GIMP and Krita, running on Linux, which have been shared freely. Those artworks are hosted for free on furry websites, like SoFurry, which are not run commercially but out of love and relying on the donations of their users. Every artwork is a product of a whole community, of a whole society. Can we now point to our artwork and say, "This is mine and you may not benefit from it in the same manner that I have benefited from so many others"?
Concluding words
When it comes down to it, I believe that AI art programs will make it possible for a lot more people (who don't have time to develop advanced art skills) to take their creativity and express it. And that is very much in-line with furry values.
Everyone gets to create.— Mary E. Lowd (@Ryffnah) August 17, 2022
At this point, I hope to have achieved two things. I hope to have helped readers better understand how AI functions and to have corrected a common misconception about how AI functions. To me, this is important because it is difficult to have a constructive discussion about a technology if we don't understand how it works. Secondly, I hope to have presented a different way of seeing the AI debate and where the world could go. The social effects of AI art are still to be seen but I hope that they are not seen purely as an issue for artists. Instead, I see the AI debate as being one part of a larger discussion about idea sharing which involves issues like free software and open access to scientific results.
AI will change things and, perhaps, some ways of doing things in the past will no longer be viable. We should try to limit the negative effects on people but we must not overlook the positives. As someone who has tried drawing at various points, I respect the skill that artists have spent time and effort to develop, but artists are not the only people who have stories and images in their heads. How many creative talents are currently excluded because they lack the time to gain those skills? For whatever flaws it may have, AI also offers the opportunity for a more inclusive world, where art creation is open to more people and more ideas can be shared. I hope that is where we are headed.
About the author
Rakuen Growlithe — read stories — contact (login required)a scientist and Growlithe, interested in science, writing, pokemon and gaming
I'm a South African fur, originally from Cape Town, who spends most of his time in Europe. I'm interested in all sorts of things, particularly science, furry and some naughtier things too!
Comments
Just as an aside, Aaron Swartz's work also directly impacted Flayrah. Swartz was one of the people who developed the RSS standards which are used by Flayrah, and countless other websites, to push content updates to people. In addition, he worked with the Creative Commons team who developed the Creative Commons licences which are the default for Flayrah submissions.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
This is not as bad as some of your positions, Rakuen, but your arguments are in line with your general ethos of "whatever gets me more stuff is correct." I mean, whatever your individual argument for something is, you seem to come down on the side that advantages you every time. Your free speech arguments allow you to say anything without consequence, your cultural appropriation arguments allows you to take anything without consequence, which this argument does as well.
I'm not even saying your arguments are incorrect, individually. But, taken as a whole, they add up to meaning, at the very least, maybe you shouldn't be the one making them.
Also, kind of arguing a side point to your main thrust (but I would argue you're arguing a side point to the main problem I have with AI art), but Mary Lowd's argument that "Everyone gets to create." is just bad, because when you give people the ability to create "who don't have time to develop advanced art skills", they create bad art. Like, okay, you know there's a phrase in customer service, it's terrible, it goes "The customer is always right." Yeah, sure, to their faces. You smile and nod and let the idiot feel like they did something, and then you go back in the back and do the job correctly like you were trained to and they weren't.
And, have you ever worked in a kitchen? Like, seriously, how do people mess up ordering a simple sandwich or pizza so, so badly, again and again? There was a semi-viral article recently about modern day manners, and it advised that your hamburger should not be a salad. And you'd be surprised how often that's the norm. You don't like onions? Cool, neither do I. Add some jalapenos? Ooh, getting spicy, nice! And can I ... nope, three strikes, you're out. You've just been rude, but furthermore, probably whatever monstrosity you're on your way to committing won't even be very good. Okay, that kind of got off topic (obviously a sore spot for me) but also, like, people order literally dangerously all the time. Like, if the menu doesn't offer it, don't order your hamburger anything other than well done, because that means the cooks aren't trained to do anything but that, and they will kill you with undercooked meat.
The saddest story I've ever heard is my brother who puts in hard wood floors for a bunch of nouveau rich-off-oil types, and a guy ordered their most expensive wood floor, not knowing it was that expensive because a. it's actually poisonous, which means the people working on it have to take expensive precautions to make sure they and the customers aren't harmed by it, and b. the wood is actually beautifully multi-colored, with blues and greens, which makes it worth the extra expense for some people. So, the floor was installed like that, only to have the customer angrily complain that he didn't want any blue or green on his floor. So, the whole floor was redone, at even more expense, to make it all brown, but still this rare, poisonous wood, so this idiot could brag about how he got the most expensive wood when way cheaper stuff would have done the same thing without potentially poisoning him. He was paying extra for blue and green he didn't want because he didn't know what he was doing. He made a bad floor.
Of course, that's all perhaps a bit beside the point, and AI art probably won't potentially kill anyone (non-art AIs in general, I mean, I've seen enough sci-fi/horror to guess that's a possibility), and I don't really say you just plain can't make good art with it. But, when someone does finally use it in a decent way (and I've yet to see something much beyond the "huh, that's funny" level from AI at best), it's not going to be someone who, well, doesn't know what they're doing. The limitation of AI that makes it an art tool rather than an artist is that it ultimately can not make choices; to use your example, it can perhaps recognize a picture of a fox, it can create a picture of a fox that I can look at and say "yeah, that's a picture of a fox", and it can then create a second picture of a fox, but if I asked it which of the two pictures of the fox it made is better, it cannot decide. Art is subjective, sure, but at the end of the day, a person can pick out which picture of the same subject they like better, meaning, ultimately, creating "good" art with AI is more about "curation" than "creation". Basically, it is a matter of tastes, sure, but even tastes can and must be trained. I'm a movie critic; believe, I know.
Ultimately, Lowd has the argument backwards; instead training emotionless, tasteless machines to "create" art without all the effort that goes into it, wouldn't it be better to train machines to do the emotionless, tasteless jobs that take up all the time and energy real people need in order to develop advanced art skills? Like, I'm not as down on capitalism as much as furries, but that's some ultra-capitalist dystopia shit right there; machines that do all our leisure and recreation for us so we have more time to work jobs?
I guess I'll take "not as bad" as a compliment. I think your characterisation is quite unfair though. For one thing, you'll find most people hold positions which "advantage them." But you seem to imply that those positions are cynically chosen to benefit me rather than that I developed certain values and now my actions are in line with those values. Do you also question whether gay people support gay marriage because they think it's right or just because they benefit from it? It also seems like you're cherry picking examples; one of my other major positions is vegetarianism which has many daily disadvantages, e.g. I have 1 option at the cafeteria instead of the 3 or 4 my colleagues get. At some restaurants I get nothing. My belief in free software and privacy means that I choose to run Linux and avoid many common programmes. I've got Steam games that don't run on my devices and sometimes have to go through hoops where other people use tools that are simply more convenient.
I agree that people often do stupid things because they don't know what they're doing (Although I didn't quite follow what the issue was in your hamburger example.) but art is subjective. I'm sure you know that. Your tastes in films are not the same as everyone else's and that's okay. I do think art can be technically bad but that doesn't necessarily make it bad art. It's not clear to me what you think about AI art will make it bad art. There are plenty of professional artists that are doing stupid stuff that I would say is bad art (most abstract painting, taping a banana to a wall). If people can use AI to generate the scenes that they want to see and they enjoy that, how is that bad art?
You say you haven't seen AI do anything more than "huh, that's funny" but AI-generated art has literally won art competitions! (See https://edition.cnn.com/2022/09/03/tech/ai-art-fair-winner-controversy/index.html and https://www.forbes.com/sites/paultassi/2023/02/05/ai-art-accidentally-wins-desti... and https://www.thesun.co.uk/tech/21265059/sunset-photograph-wins-contest-scary-twist/) Even a couple of years back, humans couldn't tell human art from AI art and even thought AI art was better! (https://www.dazeddigital.com/artsandculture/article/36940/1/people-chose-ai-made...) In that last case, that's the sort of art that gets into shows but which I think is stupid and as much bad art as any other.
I fully agree AI can and, perhaps, should be used for boring jobs but it can also be used for art. I'm certainly not saying that we should use AI to do art so we can work and I don't think Lowd is saying that either. In fact, I don't think anyone is saying that.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
Well, good counterpoint on the vegetarianism (and I guess that might also make the hamburger metaphor even harder, though it was already just a side tangent), I suppose, but I still wonder if your point (I mostly agree with) that copyright laws are overextended would change if you actually owned any copyrights.
On the art winning AIs:
First of all, this is my "curation" argument; if it takes 900 times to do something, you've still put a lot of time developing something (and even then, it is not entirely AI created; he had to modify it to make it not suck).
Second winner ... huh. I mean, I'll call it a draw, at worst. Like, if untrained idiots are running the contest, of course they are going to make untrained idiot choices.
The third contest winner would be the strongest, but your source is, well, The Sun, and The Sun's only quoted sources come from the creators of the picture, not the judges in the contest or other experts. But, anyway, this is what he has to say about himself:
In other words, he's still put in a lot of time and effort; he just didn't shit out "creativity" one day and get an award. The two awards judged by actual experts went to artists who put a lot of effort into their art, and the one who didn't won a contest by some video game PR guys who were barely paying attention.
As for critiques of modern art, well, you described Impressionism as a "style" that was copied by other artists, and not a specific artistic movements made by a group of artists with an express goal and purpose that was both a response to and rejection of previous artistic movements and is technically the original "Modern Art." The point I am making is that bad art in traditional mediums does not excuse bad art in new mediums; rather the opposite. The purpose of art is not to look like art. It is to be art. And even your contest winners don't seem to have much to say besides "gotcha!"
Whether you reject or accept my pseudo-Marxist reading of Lowd's take, it's still baffling to me that "this is going to make creativity easy!" is a take. It's just going to be this, all over again.
Like the copyrights I have for my 90 stories here? Or my over 200 art/story submissions on various sites? I've also got scientific articles, a couple hundred personal blog posts and so on. Pretty much all of which are shared freely. The only exceptions are some early scientific papers because we didn't have funding for open access but which I try to do when possible.
I think you're setting a much higher bar for AI generated art than anyone else which makes me think you're trying to have a different discussion. Looking at the rules many furry sites are going with, they are counting it even with AI assistance for a background or something. I've actually made those same arguments as you (https://inkbunny.net/j/467389#commentid_2536207); that AI generation is a tool and people using it to make art will need to learn their own skills to get the best results.
It's not that creativity will be "easy," it's that it will be more accessible. AI can help bring ideas into existence but you still need to have the ideas and something to provide as a prompt. What Lowd and I are saying is that there are people that have creative ideas, perhaps great ideas, but lack the skills required to share those ideas with others. If someone who has a really cool story in his head but no time to write a novel or learn how to draw can use AI and prompts to bring that idea into a form which can be shared with others, then that is good.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
Okay, I'll just stop leading these with semi-veiled insults, how 'bout that?
But, anyway ...
No, it isn't.
There is no idea good enough to justify sloppy execution. Besides, we already have a place for sharing "creative" ideas without effort. It's called Twitter.
But I think the argument is kind of academic because I don't think it will help people share these "ideas" because at the end of the day people with "really cool stories in their head" are not "creative", and AI isn't going to get that story out of their head anymore than a typewriter or a word processing program has been able to in the past. At the end of the day, you're advocating for a creativity based on "ideas" while I'm advocating for a creativity based on the "creative process." The thing is that not all ideas are good; the creative process, because it requires effort, makes the artist think "is this idea worth the effort". Therefore, bad ideas are winnowed out.
But even if the "real cool story" is actually a "real cool story", it still is hurt by the AI process, because there's more to a story than just the story. AI is good at academic essays because you're supposed to sound like a boring fuck for those; but storytelling is as much about the how the story is told as it is the story itself.
Therefore, a person who thinks they are creative because they have "creative ideas" will probably get bored very quickly because they have no training as to which of their creative ideas are bad and not worth actually creating for real, so most of their output will be, and I'll be generous, mediocre at best, without the saving throw of having really good technique to bolster a weak premise, while the rare actually good idea will also be hampered by a lack of good technique, reducing it to also overall mediocre as well. This person may have a bit of initial success, due to the pure novelty of it all, but eventually this will wear off, and the eventual response to this person's "art" will be audience boredom due to the sheer mediocrity of it all, leading to boredom on the part of the creative, and the abandonment of the whole thing.
Or, put it another way, I basically have a testable hypothesis, even if we are dealing with something as subjective as "is the art good or bad?". If I'm right, AI art doesn't matter, and arguing the ethics of art scraping is beside the point. If I'm wrong, well, then my whole argument is beside the point, but AI still might fail because you lose your argument about the ethics of it all.
Also, I think you should be able to mark your own comments as spam. I got it this time, though!
I don't know, what do we all think about this?
The subject matter is stupid but I like the aesthetic.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
Nailed it!
Well, we agree anyway. I think it's an interesting counterpoint to my theory that technique could save a bad idea.
Well, now I know what pictures I'm going to be seeing hanging up in fast food joints across the country now. All the more reason to avoid fast food, lol.
So now that this has expanded I will try to keep the comment brief as AI is a huge topic now. But one point being made here is that the humans are treating the computer differently than the human would treat a human.
I will argue that this is false with two main points:
1) Humans do judge other humans on the methods they evaluate or learn from the materials in question. I know this because I borrowed a "learn to draw" book from my school when I was in primary. On the bus I opened it up and instead of trying to do things free hand I started to put the paper against the book and traced directly.
Oh man, the side eye I got from the person sitting next to me could have killed.
Even when I was not selling this for profit any self gain, I was seen as cheating for doing that.
2) Humans do judge other humans when they use too much of their "inspiration" in their own original pieces. You note this "bug" in the first part where clearly the AI didn't have a robust sample set and so basically almost traced (as I did in the above example). But humans have done this too.
The most infamous example in the music world is the beginning of Vanilla Ice's "Ice Ice Baby" versus Queen's "Under Pressure".
Vanilla Ice settled out of court with Queen's estate when the lawsuit was pushed forth.
----
Overall I think your viewpoint is in a way, optimistic. Who do you think the AI is going to end up working for? The regular guy on the street or the multi-billion dollar corporations who can afford to feed the machine and hire a team of lawyers to crush the works of the small time artist who even looks their way to litigate?
Now if there ends up being an AI that can compete with a team of lawyers for pennies on the dollar then perhaps us little guy will start to stand a fighting chance. But even if that was developed, do you think the big corpos are going to go quietly into said night? They created this tangled web of litigation to protect their ass(ets). Sunk cost fallacy is a hell of a drug. It's the main foundation for why the Drug War lasted so long.
On point 1, there is sometimes a weird demonisation of tracing. I agree tracing and trying to pass that off as your own work is bad (it's plagiarism), I don't think it's a problem as a learning tool. From what I understand there are mixed feelings about it in the art world as well. I think the main question is how effective it is as a learning tool. Tracing is a part of the art process though and has been used by many famous painters; for example Johannes Vermeer (see https://en.m.wikipedia.org/wiki/Hockney%E2%80%93Falco_thesis and https://www.theguardian.com/artanddesign/2023/feb/10/illuminating-johannes-verme...). Tracing is also the key part of the rotoscoping animation technique (https://en.m.wikipedia.org/wiki/Rotoscoping). I'd say those people that are just saying "tracing is bad" with no follow up or nuance just don't know enough to have a conversation on the topic.
Point 2, I think that's a very cherry picked example. I don't know about the controversy there but incorporating inspirations into works is not unusual and can be done and celebrated, recent musical examples that spring to my mind are Nightwish's The Greatest Show on Earth ("The song features short excerpts from Dies Irae, Minuet in G major by Christian Petzold, Toccata and Fugue in D minor, BWV 565 by Johann Sebastian Bach and Enter Sandman by Metallica,accompanied with a Tibetan chant sung by male parts of the choir which are probably references to the evolution of music and arts as part of evolution of the human race." https://en.wikipedia.org/wiki/Endless_Forms_Most_Beautiful_(album)#Songs) or Sabaton adapting Tubular Bells in the opening to The Christmas Truce (https://youtu.be/HPdHkHslFIU?t=57). References are also common in animations and usually celebrated as Easter eggs. I recently watched Bagi, The Monster of Mighty Nature and there was a clear reference to Disney's Snow White, adapting the "Mirror mirror" scene and even including the box for bringing back a heart.
Perhaps I'm overly optimistic, that's why I also push for free software, so people are in control of what they are using. The problem with AI is that the training data and processing power is generally too much for individuals to work with. We will see what happens. If nothing else, more and more, the tools for people to free themselves from corporations exist.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
Well said.
How in artists’ eyes can using AI be theft and drawing fanart of a copyrighted character not be?
Not a copyright lawyer, but from my understanding, fan arts would actually be more in line with intellectual property protection then copyright.
If you were to make a new game with Mario and Luigi, then you could be found in violation of Nintendo's Intellectual property of the Mario brothers. If you share a rom of Super Mario Odyssey, then you would then be found in violation of Nintendo's copyright of the specific game.
Though, oddly enough it seems that in practice "Intellectual Property" is more of a subset of "Copyright" rather then being exclusive. Which just seems odd to me, but that's law for you.
For example, the things that Growlithe notes that some original IPs were derived from what was at first fanfiction. It is important to realize that it's not too tricky to change things so that you take a piece inspired by fan work and make it into an original IP, which works out better for the artist in the long run.
There are certain companies that will be more vicious with IP protection than others. Nintendo and Mario versus Sega and Sonic being examples of heavy contrast. Sonic fans have whole forums that they discuss making of fan games of all kinds. Mario, Pokémon, or Metroid fans don't, because drawing attention to the IP violations usually leads to cease and desists rather swiftly.
In a surprise shock twist no one could have possibly seen coming, AI fiction isn't being created by people who like fiction; it's another automated grift that is being mindlessly spammed at publishers to the point they're having to shut down. Even if this is not grifters grifting, and this massive influx is genuinely people hoping to use to AI to create an interesting story, I mean, this is what I mean when I say people who don't understand the creative process don't understand the creative process. They can't even figure out that even if AI can write a decent story, a real person has to be able to find it.
Like, once again, the copyright thing is so beside the point; the problem with AI is not even that's it's so easy to use that even idiots can use it to make art. In fact, it's not. The problem with AI is that idiots believe it is.
For fuck's sake, when your use of robots is so boring science fiction publishers are all like, "naw, get this shit away from me," maybe it's just not that good.
Need an AI to sort through and oust the AI submissions.
Mutually assured creative deconstruction
This is assuming that any of the AI submissions are good enough an actual human reader would read them and think "yeah, this is worth publishing", which is a major assumption. I mean, the guy who won the art contest in Rakuen's first example rejected, by his estimation, 900 duds, giving the AI a success rate of 0.1% (and he also further edited the picture himself). And that was a static image, while short stories are sustained works, meaning the 0.1% chance has to occur repeatedly.
The problem isn't recognizing the AI writing, because whether a human wrote unpublishable crap or an AI wrote unpublishable crap, it's still not getting published; the problem is that, once again, these people don't seem to realize that reading a submission takes time. Like, if they sent in one AI story at a time, LIKE A NORMAL PERSON, and waited until they were rejected before trying again, I'm sure these people would still annoy the fuck out publishers, but since they went with the spam model (which, note, the even the static art contest guys were smart enough not to do), they're ruining it for everyone.
Like, I suppose the plan is to get an AI story published, and then use that as a selling point ("Look, our AI fooled a fiction editor!"), but this strategy is just stupid for doing that. Like, they don't understand how publishing works. Do they not understand people read submissions? And even if they're accepted, they're often edited? Which means AI will not be the sole author, wherever you stand on the "copyright/scraping" debate. A human will be involved.
There are ways to mitigate spam. Spam has been a thing since the forum and email web 1.0 days.
When I'm saying having an AI to evaluate a written work, then it would sort the spam as spam. It could probably recognize AI tropes and start to evaluate it which were human written.
Perhaps requiring a cover letter for the entry could assist the workflow. That way there is something to quickly assess if it's something you are looking for before you dive in and read 20,000 words.
Having an automated email that will respond and inform the person a timeline of when it'd be evaluated.
If the concept of spam is enough to cause a publisher to throw their hands up, there was probably something else wrong with the publisher to begin with. The bigger threat to publishers has been self-publishing these days. So if anything, the reader should be wary of self-published items and authors are going to have to put themselves out there via video and other means to have their audience know they are really real.
Of course, at this point it's possible to make fake but realistic video as well.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
Let's not pretend that people have not always tried to make a quick buck nor that there has never been a flood of poor quality stories or animations which exist solely to ride a trend for profit. I'm sure you're aware that for every successful Disney or Pixar film there are a bunch of cheap rip-offs that presumably make their money by tricking people into thinking they're buying the original film. AI is neither the first nor the last tool which will be misused by the greedy.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
You know what, my sources are obviously one-sided, and I'd like to see what the other side thinks they're doing, actually.
Like, are they really this stupid, and think this will work? And is it just about money, or are there "true believers" involved? Or is this actively malicious (there have been some speculation, as the main publisher complaining was vocally anti-AI)?
I wonder if this started out similar to the AI art contests, one submission at a time, maybe even curated, but the mediocre results in a more challenging medium meant they were always rejected (as mediocre, not AI) before and now they're in desperation mode.
Hmm. I want to get better at drawing. I know what I'll do! Instead of drawing, I'll start tracing AI composited pictures. 500 IQ move!
AI: Good luck tracing my work with those 8 digit monstrosities that are attached to your arms.
Furry as a community would perish if AI was allowed to continue rampaging havoc. The whole of the evolution of the furry movement was reliant on the creative expression of marginalised groups. Luddites such as the Socialists warned about the "great replacement" e.g. in Soviet films like the "Loss of Sensation", while nazi films promoted the replacement of human workers and artists through mechanisation. AI leeching off of marginalised groups is not far off how white cishet people had leeched of the black and LGBTQIA+ movements in the music industry. There's a lack of compensation and AI ideologists, being capitalistic consumerists, want more and more instead of valuing human life and labor rights. The world is increasingly seeing state reliance on corporations (aka state-corporate fasces mergers, fascism). It is a libertarian crackpot far-right view that AI ideologists hold. All that AI does is make sure everyone who works hard suffers and everyone who leeches and doesn't bother essentially gets all they want. That is not just unfair, it's a warrant for a third world war, and is the reason why many countries went to war in the last world wars. Labourers won't stand for that, nor will they stand for the "superior" AI/transhuman race aka the new-man myth. Fascists should not get to define what the future is for all humans or take bullyish positions over others, especially when it harms minorities and the most important person of all, the communitarian individual. Not only is it undemocratic, it's a form of ideological terrorism to attack the arts and its humanities.
Furry would survive as a force without marginalized groups. Its core 'identity politics' is a focus on anthropomorphic animals, which can be appreciated across the political and social spectrum. Where necessary, motifs of such groups may be adopted as premium products to allow the wealthy to represent themselves as supporting them. I see no barrier to AI doing the same.
Historically, labour replacement has been largely successful as a practice. Indeed, in the UK organized labour finds itself almost without a political party given the drift of Labour away from the left in search of a plurality, which will be interesting come the next general election. Recent strikes have had some impact, but mostly in "high value" areas - doctors and nurses. And these are coherently organized groups. If an artist goes on strike - or, say, refuses AI references - they are likely to be replaced by others.
Realistically, if you are going to fight a war, you will need machinery to do it effectively; the means of its production are not held by the masses, nor do they have the technical skills to operate them - so such uprisings are likely to meet the same fate as the Luddites, execution and penal transportation. Moreover, it is hard to "break" a trivially-copyable AI that can be deployed worldwide, out of the reach of rioting artists.
That's nice, but what's wrong with the much more simpler argument: AI art is bad because it's shitty, boring art made by shitty boring people?
It's not that scary; art is not necessary, people only do it because they want to. Like, sure, we can make test tube babies, but people still like to fuck, you know? People who think they can shortcut the creative process are completely missing the point, because the creative process is the point, not the finished product. Hell, thinking of the fruits of the creative process as product (or content or whatever) is the problem, not really AI. People who want to do things will do things, and by virtue of actually doing the thing (How do you get to Carnegie Hall? Practice, practice, practice!), will make better art; people looking for a shortcut between "idea" and "thing (they can sell)" aren't doing the work, so they not only make a worse finished "product", they are also the least likely to realize their "product" sucks. Appropriately, by treating "art" as "product", they create both bad art and bad product.
This is also a rebuttal of GR, because labor replacement may be largely successful, but what we're arguing here isn't labor. It's what you do when you're not laboring. I'm still just boggling at this idea that anyone thinks using technology to do our leisure activities for us is an idea that works. It's like an episode of the fucking Jetsons where Rosie the fucking robot maid paints in the background while Judy Jetson vacuums; you make the fucking robot vacuum while you fucking paint. That's what the fucking robot's for! Ironically, your pointing out "the means of its production are not held by the masses, nor do they have the technical skills to operate them" is the argument I'm making, though in this case "it" is "art", not "machinery". Like I keep saying, the people who want to make "art" with machinery don't have the technical skills to create art. Either they're looking to subvert the creative process in order to sell a product (in which case they're boring and evil, and will make bad art), or they're looking to subvert the creative process in order to not have to do the work to develop the technical skills necessary (in which case they're boring and lazy, and will make bad art).
Which, unfortunately, doesn't make your point in the first paragraph any less true. I don't know if you actually meant it as an accusation that a lot of furry art is "boring, bad and lazy" and often "product first", but, if you did, fair. And, oh boy, I can be cynical as shit and say, "boring, bad and lazy" has never stopped the public at large from adopting it. But, you know what, let's end on a positive note; in general, while AI has been able to "create" (when properly curated) interesting static images, they have not been very good at sustained creative works. What I'm saying is writers are probably fine. Even, say, painters are probably okay, because, as I keep pointing out, most of the value there is the process (I mean, if photography couldn't kill that artform, what makes us think fucking AI will?). Graphic designers and illustrators might be in trouble, maybe? Going more furry, while, due to being more along the "illustrator" mode of art, e621 has a grand total of 22 pieces tagged as "AI generated", so I'm just not seeing a lot of interest. Like, seriously, I'm a guy who routinely puts in the same couple of tags in the search bar just about every day; shit might as well be procedurally generated, but it's not. I'm just not that worried about it.
You're on the wrong site - the AI art goes on e6ai.net. Over 2,000 pieces so far. Don't know how much traffic it gets by comparison but I imagine not much yet.
Furry art isn't just leisure. It's work for furry artists. It is desired by furry commissioners as a product; not just to make themselves, but to represent themselves to others (e.g. for con badges or online roleplay). What you seem to be saying is equivalent to "there's no point in making films/games as a product, for consumption, the fun is in making them." Which may be true in some cases, but I can assure you that games at least are enjoyable just to have and share with friends and a lot of work for those who make them, too.
AI art is great if you want to create pictures (optionally sharing them with others) without having the skills or physical ability to create them. It may still take different skills and time to refine it to what you want, but it may take far less time to acquire them and create a palatable result.
Work is ongoing on making sustained/persistent generation, it's totally not there yet in video but I can also see it getting a lot better. To me it seems quite feasible to go from one dimension to another.
Well, AI is way better at drawing genitalia than it is hands. It's even maybe a bit better at hair and fur, actually. I'll admit the style AI is good at is appealing to me, and it's does that style well, but it lacks the variety of e621. Also, and you probably won't get the reference, but either the AI has learned to start Rob Liefeld-ing hands, or at the very least the human curators have. Lots of poses with hands behind the body, or just literally cropped out by the edge of the image.
Of course, this might still be a bit damning of furry "art" in general; we do make a lot of medium body shots with little to no actual artistry beyond some posing and the basic technique. You'll notice once again that there's not a lot of AI generated art that requires characters to do something. I don't see a lot of comics, for instance. If anything, this might be the real "revolution" of AI. Maybe we'll all realize, actually, how boring and lazy a lot of even really talented furry art is, if a robot can do it. And this is kind of my problem with the "furry art isn't just leisure." It's not that you're wrong, factually; a lot of furries do financially depend on making the same basic picture again and again as work, and AI could potentially be devastating for them. But, what I'm saying is this kind of thinking, this "furry art as product", makes furry art a factory assembly line product that appeals to the lowest common denominator. Like, we like to talk about "furry creativity", but really, a furry artist who actually does it as a full time job is just as beholden to their customer base as Disney is. Sure, if you work for Disney, your opportunities to draw graphic sexual content is limited, but in furry, you are kind of required to. But let's not even talk about porn, which can open its own can of worms, but, if you want to be a furry artist, you're going to have to draw a fox. Probably that's not a big sacrifice, but even if you just don't care about foxes, the time you spend drawing foxes your customers demand you draw is not time you're drawing the animals you actually want to draw (like an echidna or something; do people like echidnas?).
This statement, "I can assure you that games at least are enjoyable just to have and share with friends and a lot of work for those who make them, too", is muddying the waters a bit. There is a difference between sharing and selling the end result of the creative process. Yes, of course, "sharing" is a part of the creative process (perhaps the hardest part of all!), and also yes, there is real work required in the production of art (that's literally what I'm saying). It's the automatic commodification of this "sharing" that is the problem here. Like, obviously, if someone wants to pay you money for your "art", that's great, but what I'm saying is that a lot of the pro-AI people are only interested in getting paid. Like, the art should have value beyond just what you can get for it, and I think while most pro-AI arguments can be boiled down to the antithesis of this, I'm finding a lot of anti-AI arguments are, whether purposeful or not, also boiling art down to just an economic transaction. Of course, it's more complicated than "capitalism bad" or "if you get paid for your art, it's no longer art", I mean, it's so expensive to make a movie, for example, that it kind of has to be a product in order to pay for itself. I mean, maybe we'd be in a better place if we had all decided, as a species, that art and commerce should not mix, but that ship has long ago sailed. Now we just need to remember they are "mixed", not "inseperable."
I have no beef with a hobbyist AI "artist" who just thinks AI is neat, and can maybe make a pretty picture they want to share every once in a while if they take the time to look for it. There may be genuine ethical concerns about "scraping", but a genuine AI "fan" should actually care as much about those ethical concerns as anyone else, if not more so, and work on it. The fact that most proponents of AI response to these concerns have been dismissive tells me they may not be very good, you know, artists. A good artist responds to criticism; they don't say "well, you just don't get it."
I think an argument about AI replacing film/game creators may further not be productive between us, because I'm still not convinced AI can make good games/films. So, where artforms like movies/video games exist on the commerce/art spectrum is kind of beside the point. Alternatively, procedurally created content and AI bots already are a thing in games; but they still exist inside frameworks created by people. Likewise, AI tools probably are already used in VFX work and some CGI animated movies, and those are the film industries "workers" who have consistently been most exploited, but I guess I'm arguing that AI isn't dangerous to them because they're already screwed, which isn't a very hopeful position.
Like, honestly, if an AI could make an entire movie by itself, and I'd actually enjoy it, at that point, it's no longer a "tool" to create products; I think at that point it could be considered a person in its own right. And then we've opened up a whole new discussion.
Come on, this is mostly romantic nonsense.
Yes, art is a product. Usually it's only the person creating something that cares about the creative process. If I commission a picture of my fursona, I don't care about the artist's process, I care about getting a picture that fits what I have in mind. If I listen to music or watch a film, I don't care how people got together to make it, it has to be entertaining. There is a process to everything, even mundane things like fixing a car or growing crops. There's no reason appreciation of a process should be limited to art; Buddhist monks will tell you all about that.
I feel you're also doing a bit of bait and switch. AI isn't replacing our leisure activities. When AI is used to do art, it's used to create those finished products. That's what you want, the process is often an annoyance. To use the oft-repeated quote, "I don’t enjoy writing. I enjoy having written." (https://quoteinvestigator.com/2014/10/18/on-writing/) There's absolutely nothing about AI that stops people doing whatever creative pursuits they want for leisure. It doesn't even stop them doing it as a business.
The fear that AI will lead to just bad art is also unlikely. That is premised on most human-produced art being good. It's not. A lot of art, music, film etc just sucks and its made by people that just suck at making it. You've reviewed plenty of films, were they all good? Obviously not or there'd be no reason to bother reviewing them. The human aspect of art is overrated. People can find beauty and meaning in looking at the view of a mountain or looking at the stars. None of those are made with conscious intent or by any artistic process, just deterministic, physical processes. If there is beauty in those, there will also be beauty in AI generated content.
"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."
~John Stuart Mill~
Well, I think what we're dealing with, romanctic are not, is basically a grift; whether or not I'm being romanctic or not doesn't really matter, because even if all that matters is the finished product, well, I've seen the finished product. It's not that good. I'm just theorizing on why the finished product is not that good.
Like, that's really the thing that's bugging me; it's not only the pro-AI crowd assuming that AI "art" is "revolutionary". The anti-AI "art" crowd is also carrying a lot of water for their enemies, because a lot of their arguments just seem to assume AI "art" is also good. Like, the grift Green Reaper repeated is the idea that artists will be forced to use AI because anybody who doesn't will be at a disadvantage, but I'm arguing, will they? I'm going so far as to say that's not an argument, that's a sales tactic. It's how pay-to-win player vs. player video games work; if you don't pay for in game advantages, you will lose to those who do. But the difference between AI and pay-to-win video games is that paying for video game upgrades gives you a provable advantage; no one has proven to my satisfaction that AI generated "art" is actually advantageous.
Furthermore, the second part of the grift from those saling AI is that you need to get in now, because, oh, it's just in its infancy, just wait until we really start cooking, it's going to get so much better, you guys, you just don't even know. And my response is, once again, will it, though? Like, I'm definitely not being romanctic here; I'm thinking like a god damn investor, and I'm saying "your current product is shoddy, and I'm not interested in waiting for you to get your shit together, either." There's some irony there; on one hand, those pushing AI are appealing to the lazy impulses of those who want skip the creative process, while at the same time asking us to respect their creative process! I guess I don't like creating smarter AIs, I like having created smarter AIs, huh?
The whole thing about this argument, both sides (and, yes, I'm doing a "both sides", shut up), is that it's just not that special. Like, it's neither a revolutionary leap forward in technology, nor an existential crisis in art. Like, I really was kind of impressed with the stuff Green Reaper linked to on e6ai (though, off topic note, maybe he should have put a NSFW warning on that link, now that I think about it), but impressed in a "yeah, I could probably jack off to that" way, not a "THIS CHANGES EVERYTHING" way. You can procedurally generate furry smut. Cool. It's a fucking gimmick. That's all. That's my honest opinion.
I've seen too much "revolutionary" technology that got people excited for a couple years at a time ... and just aren't around anymore. You remember the Nintendo Wii's codename before it was announced was "Revolution". Those motion controls caused a lot of curiosity, and a lot of positive press, and did make Nintendo a shit-ton of money. But when's the last time you actually played a game with motion controls? Like I remember when the console launched, one of the launch titles was Marvel Ultimate Alliance, which was on multiple consoles, but the Wii version was specifically given a push, with television spots showing a kid making the Spider-Man hand gesture to make Spider-Man in the game shoot his webs. Which is, first of all, basically false advertising. That's not how you made Spider-Man shoot webs; you basically pushed a button, and the motion controls were more like you wiggled the "nunchuk" attachment a little to open doors. Second of all, when Marvel Ultimate Alliance 2 came out a couple years later, the Wii version didn't even have any motion controls, even the silly "wiggle the nunchuk to open doors" mechanics. A sequel to a game sold on a console sold on motion controls didn't even use them anymore because nobody cared about the gimmick more than a few months. The Wii was still a success because it was Nintendo, and they may have sold the console on a gimmick, but they're Nintendo, and you know their first party games were going to be high quality, gimmick or not.
And the second part of the grift was there, too. That was how the gimmick of the Wii was sold; first of all, "it's revolutionary", and second of all, "yeah, okay, really we just have a neat bowling simulator right now, but just imagine what we'll do with a bit more time later on!" And then they didn't actually do that, because they didn't need to. Turns out, yeah, people do want to swing Link's sword, but they've always been able to swing Link's sword just by pushing a button, and that works. Like, that was my question when I saw the Wii controls; how are they going to make Super Smash Bros. work with motion controls? Turns out, they had absolutely no intention of making a Smash game with motion controls, and they had GameCube controller ports available from the console launch, because they obviously knew it was a gimmick. And I could go on about Second Life and the Segway and other examples of "this is revolutionary" new technologies that were sold on their very "revolutionary-ness" and "wait a minute, wait a minute, you ain't seen nothing yet!", and I'm getting deja vu, because I still ain't seen nothing yet.
Like, Green Reaper is saying that "Work is ongoing on making sustained/persistent generation, it's totally not there yet in video but I can also see it getting a lot better. To me it seems quite feasible to go from one dimension to another." but, I'm saying, historically, no. I don't think it's feasible, but furthermore, I don't even think they're working on it. Why the fuck should they? I can't fucking trust Nintendo to follow through, why should I believe the "I don't want to write, I want to have written" people?
Also, like, we're talking about art, Rakuen.
Like, you're supposed to get romanctic about it.
AI made this horrifying picture of me as a cowboy with magnificent eyebrows.
Ah look is that your fake imaginary boyfriend next to you (clearly you had to use your l33t photoshop skillz to mask out your face)
Years before AI image generators became the focus of artist's outrage, I could go to any of several image archive sites and find more furry art than I could possibly view in my remaining lifetime, all of it free to view. So having free art to view for those of us who are good with that can't be the issue. Possibly the ability to get a certain amount of customization in said images without paying an artist could be irritating. It's been useful being able to generate images for my D&D characters as needed. The images are good enough for that. The problem I think is they are good enough for a lot of other things as well. Need an image of a wolf howling at the moon for use on a t-shirt, well now you don't have to pay anyone or (just as common) steal an image outright. No copyright issues or limitations to deal with. Now for a truely special and well done piece of art, human artists are and will remain the go-to, but that's not most artist's bread and butter. How many artists are there selling $50 character drawings? Sure they are decently done, but without extreme detail added, image generators can do those now just as well in some cases. The bar for the level of artist you have to be to earn a living at it has gone up significantly in the digital domain. The generators will continue getting better. So digital only artists are going to have to learn to compete with them, or use them as tools for hybrid art pieces (which can be amazing). Alternatively, put away the mouse and stylus and pick up paint brushes, markers, pastels, etc and switch to physical media. So far AI's can't do anything with those.
Post new comment