The Big Battle About AI Art? Which Side Do You Favor?
It’s not news that AI art can radically upend sectors and make people wonder what humanity’s place is in places where it’s being used. But art and creative expression, which we frequently consider the exclusive and sacred domain of what some call the soul, have never before been so directly impacted by technology. Most people worry about what would happen if artistic minds capable of creating works of unparalleled nuance, complexity, and depth may be trivialized by a little chip.
But now, in the middle of that existential concern. Artificial in AI art generators essentially unlocked Pandora’s Box. But it is possible that a misunderstanding of how they operate is one of the key reasons this type of art and the instruments related to it have sparked such a passionate response among artists and art fans.
So, how do programs that create art using AI work?
Technically speaking, Generative Adversarial Networks are required for AI image production (GAN). This kind of network actually uses two neural networks—one to construct the image and the other to assess how closely the image resembles the original using online photographs as a reference. The initial AI “learns” from this feedback and sends back an image to be utilized in the subsequent scoring round once the second network has created a score for the accuracy of the image. Additionally, prompt-based image production was created by fusing language processing models with artificial neural networks (ANNs) made to generate images from text input.
MidJourney, DALL-E, and Stable Diffusion are the top three competitors in the AI-generated art game. Imagen and Parti were developed by Google researchers, but they haven’t been made available to the public in part because of worries about their uses. David Holz founded a company with the same name that produced MidJourney. Elon Musk financed OpenAI and produced DALL-E, and Emad Mostaque, the founder and CEO of Stability AI, created the open-source Stable Diffusion. In terms of output, MidJourney’s photographs frequently have a more illustration- and paint-like look, Stable Diffusion frequently veers into a kind of photorealistic surrealism, and DALL-E appears to be able to straddle both of those aesthetics.
Unpacking the concept of “prompt-based” art is worthwhile. AI art is a collaborative process that blends computer and human input to get the desired result. For instance, a user can enter a string of words on MidJourney and get four visual outputs in the form of a grid that roughly represent the original concept. From then, users can iterate on those outputs indefinitely, nipping them in the conceptual bud, or upscale, refine, and tweak the outputs in an endless round of co-creational tennis. Additionally, the number of potential generative outputs of these programs approaches infinity before a user even starts to iterate on the original result because the language has the potential to convey ideas and concepts in an infinite number of ways.
Misconceptions about how these systems operate are common, maybe as a result of their technical sophistication and opaque nature. The criticism that these systems “smash” together previously existing works of art to create something new is one of the more frequent ones that proponents of AI art find themselves fighting back against.
According to Black Label Art Cult (BLAC), a proponent of AI art and a member of the pro-AI art organization AI Infused Art, the assumption that AI art steals works of art from humans is the largest problem with the narrative surrounding it, which is untrue. “Sometimes, people believe that these systems take already-existing art, put it in a bucket, piece it together, and then market the finished product as an NFT. It gives a false impression of what truly occurs in these programs.
You must train an AI model on a vast amount of data in order for it to understand how to perform the intended function. We know that the AI models used in these prompt-based algorithms are trained using trillions of parameters and trillions of photos, but the majority of the companies that developed them have not yet provided much technical information about how they created them. For instance, Stable Diffusion is trained using a core set of more than 2.3 billion internet-scraped image and text tag pairs.
The word “trained” is crucial here. These applications don’t access a sizable image database from which to extract image fragments for new works of art. They have mastered the art of connecting text to specific graphic elements.
AI artist and a key figure in the AI art movement, Claire Silver, said, “If you ask it for a human, [the software] knows that humans have two hands with five fingers on each hand.” It is aware that fingers are lengthy, cylindric, and bone-filled. It is aware that bones behave and look in this way. So, using what it has learned to create something new, it “imagines” everything you asked for.
“c l a i r e” by the amazing @ClaireSilver12
Truly one of my all-time favorites
AI art is filled with controversy, which is why I love it so much. Because in the battle of Artists v Society on what defines art, the Artists are undefeated
This battle won’t be any different pic.twitter.com/LWOGiri3NK
— batsoupyum (@batsoupyum) November 20, 2022
Silver is a loud proponent of AI art as a part of a new creative revolution that allows both established artists and individuals who aren’t very talented at producing visual art to express themselves through the arts. She also organizes well-known AI art contests on Twitter; the most recent one ended at the end of October. In the 18 days before the competition ended, thousands of people contributed artwork, and the winning pieces were shown at the imnotArt gallery in Chicago.
Gn from Chicago ❤️ pic.twitter.com/J9lu8JUczj
— imnotArt🧱 (@im_not_art) November 14, 2022
AI Art’s moral minefield
Some people have been thrilled by the explosion of creativity and art that has come with the introduction of prompt-based image programs, while others have been disgusted by it. The possibility of these programs causing philosophical vertigo has people both excited and terrified, but there are still more practical concerns around ownership, fair use, and deep fakes.
According to MidJourney’s terms of service, anyone who purchases a member’s license is permitted to use the photos they produce however they see fit, including for financial gain. These programs are now being used by artists to aid in the creation of their NFT collections.
However, the problems with what people can do with these photographs are simply a portion of the moral minefield they’ve contributed to. The potential for photorealistic results to cause significant problems with the use of deep fakes to either extort people or distribute false information online is quite real.
Both originality and plagiarism
The issue of overt artist plagiarization also exists, as demonstrated by the Reddit user who admired Chadeisson’s aesthetic. A quick peek around MidJourney’s Discord reveals that the person who posted the aforementioned AI model is far from an isolated instance; others are also developing and refining that artist’s approach. But if at all, how big of an ethical violation is this second-order iteration?
Regarding the hazy distinction between inspiration and plagiarism, Claire noted that artists have always drawn inspiration from other artists and used them as sources for master study. It’s nothing new, that. There is also the movement known as a transformative use, which involves collage artists who really take identifiable pieces and utilize them in new ways.
Silver urges users to use caution and consideration when using generative AI art tools to avoid blatantly plagiarizing the work of other artists. When working in whatever media, that is a professional and social courtesy you should extend to any artist. In the end, though, she thinks that even in instances where overt imitation or plagiarism occurs, such work is likely to act as a catalyst to identify the original creator.
The fact that there is no significant way to prove whether or not an artist’s creation was used to create the models that millions of people are currently using raises another ethical issue with these systems. Stable Diffusion does not reject copyrighted art when training its model on its two billion+ image dataset. Similarly to this, when questioned by Forbes in a recent interview if the business obtains permission from the artists whose photographs it has used to train its program, MidJourney’s CEO David Holz only stated that there was no practical way to do so.
Holz emphasized, “No. “It’s difficult to obtain one hundred million photos and determine their source. It would be cool if photographs contained metadata identifying the copyright holder or something similar. There is no method to locate a photo on the Internet, instantly identify its owner, and then take any action to validate it.
It’s an honest, though unsettling, point. These programs’ rapid development came before the clear requirement for a digital compensatory infrastructure that the programs themselves have developed. On the user’s end, there is currently no legal requirement that they acknowledge the fact that they used an artist’s production as inspiration for their own works. Holz has acknowledged that artists may one day be able to choose not to have their names included in prompts, but there is no guarantee of this. MidJourney is only one of several programs that will have to deal with the copyright issue, even if it does happen.
Is AI Art the new ability?
Because AI-generated art programs are so well-liked, there are now entire markets devoted to buying and selling ideas for fresh artwork. According to Silver, it may be a sign of a trend that people would reconsider the value of artistic skill that the thoughts and ideas these programs require have become so highly valued.
Another worry voiced by people leery of such AI tools is that the technology will evolve to the point that media firms will end up completely excluding artists from the creative process, erasing a job class of already mistreated and disenfranchised creative professionals.
Control over the AI art movement
The tale went viral after Jason Allen used an AI-generated piece to take first place in a digital arts competition at the Colorado State Fair earlier this year. The in question piece of art became the center of the entire discussion on the ethics and soul-seeking that AI art has produced. And while some artists, like Silver, think it would be wise for anyone entering a contest like this to state up front that their work was produced using these programs, they shouldn’t be embarrassed to do so.
The border will become so hazy, according to Silver, that he believes people will change their minds. “Photoshop currently has a lot of AI incorporated into it, including neural filters and various tools. When people employ AI in their digital work, they don’t refer to it as AI. I believe it is prudent to identify work produced with AI as such for the time being. Furthermore, I don’t believe that [the AI label] should be an embarrassment or a sign of weakness. I take satisfaction in the fact that you are at the forefront of this burgeoning movement.
Similarly, BLAC thinks there are many good reasons to embrace these technologies rather than vilify them. Along with other artists and AI aficionados AmliArt and illustrate, the artist assisted in organizing two AI competitions with the topic “expression with purpose.” As a result, the artist has seen individuals use these tools in deeply meaningful ways.