I am responding to a post he made on his facebook page here, his words are in italics.
Is this the future we want for art?
AI art tools like Lensa App, Midjourney or Stable Diffusion have recently taken the world by storm.
It’s impressive what these applications are capable of but most people aren’t aware of the very worrisome implications this trend has for the future of art:
1. AI tools are algorithms that need source data to function. In the case of AI art software, the source data is a pool of millions of artworks that were created by human artists over the centuries. AI tools can only create stunning images because they have access to this vast library of art. The problem is that most artists didn’t give permission for their art to be used this way. Furthermore, contemporary art is protected by copyright laws and is used against the artists’ will to feed a machine that aims to replace them by recreating their art style. This is not only morally reprehensible but most likely illegal. It is art theft on a massive scale disguised as technological progress.
Contemporary art is protected from… being copied and reproduced. The common datasets that are in use, like LAION 5-B, are just links to images and have not been copied from the respective sites the artist uploaded to. They’re literally just indexes of images that exist on the internet, with the ALT texts found linked to those images. So the datasets themselves run afoul of no copyright, since they neither copy nor store data.
The actual process of training itself would also not fall afoul of copyright, due to transformative use (using US copyright law, due to the US location of most of companies creating the AI Generators). That is, the finished model does not store the images it was trained on, and instead what it learned from them <-this is incredibly complicated and if want to see inside the black box for yourself there’s a big list of articles and papers located here. At any rate, even IF lawyers screwed up enough to allow the idea that the training of models was not transformative enough to avoid copyright, the final results of using the model are unique enough to pass copyright muster. As much as artists hate it, style has never been copyrightable.
2. AI art threatens the livelihoods of millions of artists, designers, photographers & other creatives. These tools are built to create artistic or photorealistic images in a myriad of different styles and in an instant. “Fast art“ you might call it. No human artist can compete with a machine that can create 20 different iterations of a prompt in 5 seconds. The basic economic principles of supply and demand dictate that when the art market is flooded with cheap & easy to create AI art, the wages for artists will go down even lower — when most creatives can already barely cover their living expenses. Why should companies or individuals pay a human artist when they can get AI to create their desired outcome in the blink of an eye and for the fraction of the cost? We already see this happen in certain fields and this trend will only accelerate.
The invention of cars threatened the livelihood of the horse industry. Breeders, saddle makers, carriage makers, and all their suppliers and associated things, gone because of cars. Industrialization destroyed the livelihoods of everyone who used to make their works by hand. There’s even a still well known folk tale and song in the US called John Henry, about a steel driver racing against an automated rock drilling machine back in the 1800s. Photography destroyed the livelihood of traditional artists, especially portrait artists. Digital Art tools destroyed the livelihood of traditional artists and photographers. Technology and progress is inherently disruptive by its very nature.
3. AI tools devalue art to a mere product of algorithmic approximation and a means to serve consumerism. They mash together source material to create images that are visually pleasing but have no human soul. Art has been a form of human expression and communication for as long as we have existed. From cave paintings to impressionistic masterpieces of more recent centuries — art is a vital part of who we are! It helps us make sense of the human condition and put beauty into our world. Capitalism has the tendency to reduce things to products merely designed to generate revenue. We have seen this happen in the world of movies and video games, where the same generic themes are rehashed over and over because they are „safe“ and sell. Rarely do we see hollywood directors truly tread new paths because corporate profits are more important than having creative vision. Do we want the same thing to happen to art?
AI tools devalue art only for those who have never spent time thinking about what exactly art is. The kind of people who never understood the underlying message of postmodern artists in their deconstruction of art (‘what is it that makes an art, art? who gets to decide this?’). This point also makes the erroneous claim that the AI is ‘mashing together source material’, which is functionally the same as saying anyone who studied an artist in an art museum for a while and made works in a similar style was just ‘mashing together source material’. It’s just ignorance of how these things work, and weakens the argument as a whole. This entire point is also pointless. Why does a person make art? Is it for money or because they want to create a thing? AI art is not going to stop anyone from making art in the same way that people still make saddles, and people still paint portraits with oil paints and canvas. The demand will drop, but as long as there’s a person who wants to do it, and someone who wants it, the old ways will continue to survive. Those who wish to make it a career better start learning AI the same way they were forced to learn how to use tools like photoshop at the start of the digital art age.
4. AI art apps like Lensa encourage their users to upload photos of themselves and turn them into artistic masterpieces. This poses a huge risk to the privacy and safety of the app‘s users because their likeness is fed into the AI‘s source data and may end up being used for unwanted nefarious purposes. It has been reported that Stable Diffusion, the software Lensa App is based on, has been used by some to create p*rnographic images. Would you want your face to float around in a software that can create these kinds of images?
This is probably the only salient point in the entire list. Putting images of yourself on the internet for any reason is a security risk, and it’s up to the person to determine just how internet secure they wish to be. If you post a lot of images of yourself on any social media, this ship has already sailed.
5. AI art tools enable art thieves in completely new and dangerous ways. Some artists have already experienced this: the thieves would feed the artwork they want to steal into the AI and let it create a variation. The outcome would be very similar to the original artwork but different enough to protect the thief from a copyright lawsuit. Even entire art styles can be imitated. A digital artist called SamDoesArt recently had thieves train AI software to almost perfectly copy his art style. They even made a contest out of it, rewarding whoever got closest to Sam‘s art style. While a certain style is not protected under copyright laws, it takes artists years and years to find their unique voice and perfect it. To have it stolen like that, without a legal way to fight back, is beyond frustrating.
This point could have been used to explain why AI art is not itself illegal. specifically ‘The outcome would be very similar to the original artwork but different enough to protect the thief from copyright suit’. Artists have been doing exactly this going all the way back to the early Renaissance. There’s literally books called “Steal Like an Artist” for a reason. Art styles have always been imitated, and in some cases that imitation was a requirement -the ‘house style’ of a company, or the ‘Marvel Style’ that existed for example. Art styles cannot be stolen, as the style is not legally ‘theirs’ and never was to begin with.
6. Lastly, this AI art trend discourages people to get into the arts and develop the necessary creative skills it takes to become a professional. I myself had many aspiring creatives reach out to me, telling me how scared they were for their future and that they didn’t know if going to an art school or studying art by themselves was even worth it anymore. Choosing an artistic career is already daunting and difficult enough. Having to compete with machines that have no downtime, no self-doubt and learn at a much faster pace than humans could, will cost us entire generations of actual humans wanting to add their creative voice to society.
This same exact argument has been used in my lifetime by people I know in regards to digital art, and specifically photoshop. All that hard work and skill that goes into traditional art forms, can be negated by people churning out ‘perfect’ and ‘soulless’ art by ample use of the ‘undo’ button. There’s no depth to the texture you can see on the canvas of artists who spent layers and layers of paint trying to get a feature right, or correct a mistake. It’s soulless, and empty of life they said. Digital Art is not art. Now look where we are. And I highly disagree that it is discouraging people from getting into the arts. I think the massive and explosive growth of platforms like Midjourney say the exact opposite. People who could never before put their imagination down on any medium, or afford to pay for a commission can use the AI to express themselves in ways they would have had to spend their entire life learning how to do.
I am not against technological progress. I think AI is fascinating and can be used for good in so many ways. I just believe that we shouldn’t automate things that are so central to who we are as humans. We risk losing touch with our own humanity. Think about how much art has an influence on our everyday lives: it’s in the music we listen to on our way to work, in the design of the car we drive or in the Netflix show we watch to wind down in the evening.
He, like most people, are against technological process when it has the ability to affect him and interferes with his industry. Music, Car design, Netflix shows, are all corporate made. Artists are not given free reign over what they want to do, and for a significant part of the art industry, the artists don’t pour their soul out making someone else’s work any differently than a carpenter framing a house, or a tailor hemming clothes, and as such, the entrance of AI into those fields is also not going to make any noticeable change to the people who consume the content.