If you’ve been paying attention to the ground for the last couple of years, then you’ll be familiar with some of the arguments over AI art’s ethics and impact. Some of the tools used in creating AI art include Midjourney and Stable Diffusion. You may be curious as to why these tools are generating strong opinions on the news, social media and among your friends. Haven’t we had the “robots are going to take our jobs” discussion for decades?
These AI tools, which are publicly available, can respond to incredibly specific prompts, and then unflinchingly portray them. It’s like an artist working on a commission. They don’t care if you want a life-size painting of Mario or Luigi eating a Toad barbecued. These tools often do this for free. Many people are using tools like DALL-E to generate memeable images on social media, but others saw the commercial potential behind AI tools, and it wasn’t long before an artist entered a piece of AI-generated art (using Midjourney) into a competition — and won, causing outrage and concern for the art industry.
There are AI-generated videos, also. They are not exact. GoodHowever, AI is an emerging trend that could spell doom for many artists and developers who are worried about their livelihoods. We spoke to a handful of these creators to find out what the general consensus and mood are in the games industry towards AI art, and whether we should be worried that robots really will make us obsolete — or worried about something worse entirely.
What does the AI art community think?
Ole Ivar Rudi (Art Director at Teslagrad 2 and Teslagrad 3) feels that the AI art situation is very complicated. He tells me via Twitter DM that he is “a bit on the fence”. “On one level I see the appeal of it and find it fascinating. [but] The data sets are mostly based on unethical material, including the work and illustrations of illustrators. This worries me a lot.
It’s inherently fascinating to throw a coin into the wishing well, rub an oil lamp and ask for something.
However, he does admit that the results do have their merits. He says that on one level, he can see the appeal of it and find it super fascinating. “There’s something intrinsically interesting about throwing a penny in the wishingwell or rubbing an olive lamp and asking for something. (Conan the Barbarian riding on a lawnmower). A werewolf ordering French fries!) then you’ll get an unpredictable, distorted version of what you envisioned in your mind when you typed your prompt.
Martin Hollis is a game designer best known for his role in GoldenEye 007’s directorship. He believes that AI art has value because it can produce results that are both meaningful and useful. Just So random. He says that most of the most important images he has seen are worth their value because they are humorous. “Particular humor comes from the AI’s inability to draw or understand… For example, many AIs have difficulty drawing their hands.”
And that’s funny — in the same way Botnik’s “AI” predictive keyboard scripts are funny, because they go to places that make no sense, even if the grammar is Technically correct.
“Mario is a fictional jerk. He is a Norwegian woodworker who treats women cruelly.”
Botnik has provided an excerpt from the “Mario Wikipedia Page”
Karla Ortiz (a concept artist and award-winning designer whose clients include Marvel Entertainment, Universal Studios, Wizards of the Coast, and Universal Studios) believes that AI art might have a place. She tells me via email that she can see many very interesting applications for AI. It could be used to find references, create mood boards, or even assist with art restoration.
Ortiz is optimistic about the future of AI art, but her optimism is heavily tempered with its flaws. AI art, which draws from uncredited sources images, is Ortiz’s main problem. They are only allowed to have a role in the art industry, she states, “if they are credited source images.” [they] They were only ethically constructed using public domain works, with the express permission and compensation of artists’ files and legal purchase of photo sets. This is not the current situation.
Is AI training data infringing copyrights?
Ortiz compares the AI art forms currently in use, such as DALL-E or Midjourney to a calculator or a photo mixer. They are unable to subjectivity and make decisions solely based on their programming.
This leads to an issue at the core of algorithmically-generated art: It can only learn by copying. AI is not able to be creative on its own — you have to teach it, using a library of training data. This could be a literal collection of books that teach an AI to write or a collection of music, art, descriptions, and descriptions that teach it what is “good” or “right”.
Even AI companies are unanimous in their belief that current AI models copy copied data
Machine learning is based on the principle that more data leads to a deeper and more comprehensive understanding of “art”. The internet is the largest library we have. It is a place where ownership is rarely respected and any post without a watermark can be considered “free game” (and sometimes people remove the watermark).
The AI then uses that data to extrapolate. Ortiz says that “the software makes an unscientific guess about what an acceptable picture is, based upon the original images it’s been trained on.” Ortiz says that without strict supervision and careful selection, there will undoubtedly be copyrighted materials in the data. This is not a secret. “Although AI companies are aware that AI models currently use copyrighted data,” says Ortiz.
The creators of AI generation tools know that copyrighted media could be a problem for their training data. Ortiz cites the AI music generation tool Harmonai’s statement on the topic, in which Harmonai claims that they use no copyrighted music in their training data. This is proof that companies are well aware of this issue.
“Diffusion models are susceptible to memorization overfittingRelease of a model who was trained using copyrighted data could lead to legal problems. It was important that any copyrighted material be kept out of the training data.
In machine learning, something is “overfitted” when it sticks too rigidly to its training data — like a child reading “Tom went to the store” on the first page of a book, despite the first page being the author and publisher information, making it clear that the child has just memorised the book and doesn’t actually understand how to read yet. Ortiz explained that this means AI companies cannot avoid plagiarising artists’ work.
DALL-E’s training data, for example, is described in one of their blogs as “hundreds of millions of captioned images from the internet”, and the engineers discovered that repeated images in that data — multiple photos of the same clock at different times, for example — would lead to the results “reproducing training images verbatim.” They created an additional algorithm to “deduplication” which detects and removes similar images. This was done in order to minimize the risk. almost a quarter data being removed.
OpenAI engineers from DALL-E are not certain that they have solved the problem of “memorization” even after they’ve completed deduplication. “While deduplication is a good first step towards preventing memorization, it does not tell us everything there is to learn about why or how models like DALL·E 2 memorize training data,” they conclude at the end of the blog. It’s clear that there’s no way to stop an AI copyrighting images. OpenAI has admitted this in their “Risks and Limitations” document.
Then, who is the rightful owner?
Users can’t determine whether copyright and/or private data were used in the generation process.
The unregulated use source images raises several issues. The client-facing side is not transparent as many AI tools don’t make public their training data. Ortiz says that, even if a company has strict guidelines regarding the use of any type of copyrighted materials as a prompt for AI models, it is difficult for users to determine if private or copyright data was used in generation processes.
So who is the owner of the copyright to an AI generated image? This AI-generated picture has used an unknown number of potentially copied images to generate something unique. The debate continues. A paper titled “Who owns the copyright for AI-generated art?” was published recently.Arthur Roberts (a trademark attorney and specialist in patents and software) is unable provide a concrete answer to this question. This turns out to be at least in part because the ownership of the image is unclear — is it the creator of the software? Or the creator of the training information? The user who created the prompt
While they state that the answer will be available in the coming years, it is best to evaluate each case individually. They offer advice rather than providing answers. The terms of service and licensing agreements should be reviewed to ensure you are aware of what you can and cannot do with an AI tool.
Although we don’t yet know the answers to these questions, Roberts and Godement have made clear one thing: Law surrounding AI art and copyright owners is murky at best.
Who wins and who loses?
Aside from all the copyright issues — is AI art an actual threat to anyone’s careers in particular? It’s difficult to say. It doesn’t appear that the technology can be legally and openly used as a tool for creation. However, not everyone is so strict about legality.
Hollis considers AI in professional art creation to be a natural progression. It seems to be inevitable. [likely that] He says that the technology will only be used in “a few subdisciplines of the industry”, but that this will not be the case. Look It’s as if they were created with AI art. They therefore belong in their own category. “There is absolutely no chance of videogame makers becoming fewer – there are more people making them every year.”
A growing consensus is that we will see some job loss in the entry-level positions, at least.
Ortiz sees AI art as a threat to both concept artists and newcomers to the industry. She says that there is a growing consensus that we will experience some job losses, particularly in entry-level jobs. While people with her expertise and experience may not be directly threatened, losing junior positions could have serious repercussions for the entire industry.
Ortiz says that “these entry level jobs are pivotal for the overall health and livelihoods of our creative workforce ecosystem” and notes that losing them would make it difficult to access the industry. These entry level jobs are crucial for artists not from wealth backgrounds.
Rudi agrees that automation replacing workers is a benefit only to those who have too much money. “I feel a little uneasy considering how poor just about everyone is doing economically these days.
Ortiz says it’s worse, but at least the production lines weren’t shut down. Literally steal Workers. “Unlike past technological advancements that displaced workers, these AI technologies utilize artist’s own data to potentially displace those same artists.”
Rudi agrees, envisioning a more specific future scenario. “I am certain that I am worried about that […] Some people will happily accept a computer-generated pastiche of the artist’s style that is warts-and all computer generated, instead of hiring an artist they love for their commissions.
In fact, one particular area that AI art could feasibly be used is in creating Pokémon designs. Several AI Pokémon generators exist, from Max Woolf’s tweaked version of ruDALL-E, which you can use yourself in his Buzzfeed quiz that generates you a unique Pokémon, to Lambda Labs’ Stable Diffusion-trained generator, which lets you input any text you want — an IKEA desk, Boris Johnson, a half-finished sandwich — and it’ll turn it into a Pokémon.
You can See more the training data in the results — an arm of a Gardevoir here, the shape of a Chansey there, plus Ken Sugimori’s trademark style — which just goes to prove that AIs are not creating anything unique as much as they are image-bashing. And although a tool like this certainly wouldn’t put industry veterans like Sugimori out of work, it could replace more junior Pokémon concept designers. After all, Pokémon designs are iterative — there are always evolutions to design, or regional variants, or new forms, and taking something and tweaking it is what AI generation tools excel at.
Mass production of art that is similar to the work of another artist. […] This behavior should be considered as harmful, parasitic, and socially unacceptable.
Hollis says that “stealing” is somewhat of an art world relative term. “Is it theft for humans to learn from other artists’ work?” He asks. “We have a complex system for ethics surrounding the use of other people’s work in the art world. We have fraud at one end, which then turns into shameless imitation, plagiarism, and homage at the other. Surprising originality can be found at the other.
However, AI art does not have to be original. Hollis admits that AI art can sometimes be a bit invasive. “Naturally, when a program produces mass-produced art in the style or legacy of another artist, it must be deemed parasitic, damaging, and socially unacceptable. Otherwise, we will continue to look at these microwave meals of actual artist’s work for the short term.
Ortiz goes further and points out an egregious usage of AI technology. “Users take and degrade work of the deceased for their own purposes, and without permission or disrespecting the wishes and wishes of their families.” In the wake of Kim Jung Gi’s sudden, tragic death in October, someone entered his artwork into an AI generator to “honor” him. He asked for credit and was met with outrage by friends and family who thought it was an insult to his memory and art. You cannot, after all, replace a human with an algorithm — but that doesn’t mean that people won’t try.
We will see AI art everywhere.
Between the ethics and legality of AI art generation tools using copyrighted data in their training models, and the moral implications of what that means for a user — and, indeed, how they choose to use it — it seems like AI art will struggle to find a firm footing in the eyes of many. It doesn’t necessarily mean everyone feels the same way. Some may choose to boycott the technology. Others might view it with open suspicion.
AI art can be used to create images that are very specific and with large amounts of eyes. This is what many people consider AI art. gigantic chests, or random mash-ups of pop culture references, to garner likes on social media — and that’s all it is. This is not a systematic demise of an industry or unethical, non-consensual exploitation of artists’ works. The majority of people don’t know much about AI. They just want to be part of a new trend. AI art generation tools are easy and affordable. These people might not have asked for an artist to draw “Pikachu with a swarm bees in Picasso’s style” in the first instance.
Others, particularly those potentially affected by AI art, have mixed reactions. Some see its application as a tool for humour, others see it as a potentially helpful tool for sparking creativity — but it seems like everyone can agree that the technology leans too heavily on the side of plagiarism, although some disagree about how serious that is.
Because everyone is talking about the art, it’s impossible to argue that it’s boring.
Hollis believes it could all be a passing trend. He says, “It doesn’t really matter if AI artists’ are ‘good or ‘bad. They are fascinating. It’s hard to argue that art is boring right now, since everyone is talking about them. It will take six months before it becomes boring again, until the next technological advancement and step change. His current popularity of AI art as a hot topic is its novelty, he said. “When it stops being new, then it must survive on its merits,” he said.
Ortiz’s skeptical attitude about technology is temperated by a tiny bit of hope. “I could see very interesting use cases AI,” she says, especially in her work line, where AI art could be useful as references and mood boards. But the technology itself needs to be rebuilt from the ground up for her — and many other artists — to feel comfortable about its use. She says that these tools are very interesting. They can be ethically built, and those who profit from unethical tools must be held responsible.
Your opinion on AI art? Are you concerned that AI art could prove to be a dangerous tool? Is it a useful tool for creating creative concepts? The industry is at risk? Fun way to make silly pictures You could try something different. Comment below to share your thoughts.