The Dark Side Of A.I.: How Artificial Intelligence Is Threatening The Future Of Writing And Art
Image Source: CultureSlate and Anthony C. using Stable Diffusion
Imagine you have worked really hard on a piece of art for over fifteen years. It is your life. Despite all the tears, it also brings you great joy. You have worked hard. You have created something. And you cannot wait to share it with the world. After all those years of fine-tuning and making it your own, you get ready to release it into the world…only to have another, almost identical piece of art released just ahead of yours, completely overshadowing your hard work, and taking the world by storm. It should have been you! Where had this art come from and so fast? And then you find out the worst of it. This similar art piece did not take the creator years to work on. It did not even take them a few months to work on. It only took them minutes, seconds even, and they had stolen your thunder. How? With Artificial Intelligence technology.
This extreme case is a fear that many artists and creators have with the sudden uptake of how dangerously good Artificial Intelligence (A.I.) technology has been at creating art pieces and stories. While there are benefits to using A.I., the uptick in usage to create art in a matter of minutes is concerning to the art and writing community. The A.I. software that has been generating a lot of buzz are Midjourney, Dall-E 2, and ChatGPT.
RELATED:
A.I. writing and art software have traditionally been a joke. There are famous examples floating around the internet of people who fed basic movie plots into an A.I. generator the result was hilarious and did not make much sense. It was not coherent writing, and it clear which artwork was real and which was created by A.I. However, in the last couple of months, A.I. technology has suddenly gotten so good to the point where it is much harder to tell which is done by a human and which is A.I.-generated. The writing A.I. will actually spit out mostly coherent and semi-good plot and character- driven stories. The image A.I. actually makes fairly good works of art. Nothing completely out of the ordinary, but still good. And it is all generated by code in minutes.
The team at CultureSlate played around with ChatGPT to see what it would spit out. At first glance, the results were actually really compelling and fun stories. However, the more you read into them, the more you realized, it was still just words on a page, and something was definitely off. The Washington Post’s experiment using ChatGPT to create a children’s book like Maurice Sendak’s Where the Wild Things Are or Mo Willems’ Don’t Let the Pigeon Drive the Bus completely changed the actual message or feel of the story and in some cases, added morals that were not even there. There is no feel to the stories. There is no heart and soul. They are just words on a page. For CultureSlate’s attempts, the result would be a really great short film, but, because of the lack of feeling and the not-as-polished writing, there is still a lot of work that needs to be done.
Image Source: Anthony C. using Stable Diffusion
The same can be said of image A.I. such as Midjourney. The art is there, but there is no specific style to it. Sure, with certain images, it can actually be hard to see at first they have been created by A.I. However, if you know where to look or you look at it long enough, you will start to notice things that are telltale signs that a human artist did not create it. The major problem with art A.I., though, is that you do have to feed art to the A.I. to give it a base to work with. The issue is, it is taking a lot of these famous and non-famous artworks from the web, which is fine in some sense, but at the same time, it is actually turning out to be an ethical problem. As The Washington Post pointed out:
“The tools reproduce and remix images by existing artists who aren’t paid for, or given the chance to consent to, the use of their work. Many artists are understandably concerned that a tool that rips off their styles could be used to replace them.”
According to Stable Diffusion’s own data, they took images from over 100 sources with Pinterest being one of the top ones followed by Smugmug, Tumblr, and Blogspot. The Washington Post also pointed out that Stable Diffusion also pulled from personal blogs and amateur art sites like Flick and DeviantArt. There were also images taken from shopping sites such as Fine Art America, Redbubble, and Spotify. Of course, many did come from stock image sites such as Getty Images, which is not quite as worrisome since these are free for anyone to use for the most part. Of course, numerous images are from famous artists (most of which are not alive anymore), whose photos of their art are available on Google for anyone to use. However, this is concerning for an individual artist (or anyone who has a social media account). If you put your work out there, people can use it and not pay you. It sucks, but it does happen, and A.I. is not the first to take art or photos that are technically on the web for (mostly) free and use them in its database. It is clear that there are still a lot of moral implications and copyright issues to be explored here and they are currently being scrutinized.
In regards to copyright, what happens when you feed in a picture of Mickey Mouse, for example? Or feed in a story that has already been written and it pumps out almost the same exact story? While technically A.I. makes things different enough that at this point you can mostly get around copyright laws (for the time being that is, this is a fluid issue), it does beg the question of what happens when A.I. grows and becomes better as it inevitably will. What happens when it becomes almost imperceptible to tell which is the real Mickey Mouse drawing? How will copyright laws react to that? Some are already being pushed.
Many entry-level artists in Hollywood and various companies are expressing concern over A.I.-generated art potentially taking their jobs. While it is indeed known that A.I. art still requires additional work, much of the art could be used for entry-level work, and completely bypassing the human. The same could be said for scripts/stories, etc. Self-published authors could bypass hiring an artist for their book covers and do it themselves. It would require time, but it would save them money. This sounds great, but it now affects those artists whose jobs are now being sourced out to A.I.
RJ Palmer, who worked on Detective Pikachu and at Ubisoft, voiced his concern in an interview with Kotaku:
“Doing that kind of work for small creators is how a lot of us got our start as professional artists. So as an artist seeing this attitude grow gives me concern for the next generation of artists being able to find consistent entry-level work.”
Karla Ortiz, who also worked at Ubisoft and Marvel, said in the same interview:
“The technology is not quite there yet in terms of a finalized product. No matter how good it looks initially, it still requires professionals to fix the errors the A.I. generates. It also seems to be legally murky territory, enough to scare many major companies.”
A friend of mine has been playing around with photo A.I., using a free version of Stable Diffusion. The A.I. might not be able to figure out what a crossbow is, but other times it creates some absolutely almost perfect photos. Most of the time though, like Ortiz said, he does have to tweak it after the A.I. spits it out, or even before he sends it through the generator. This does add more work than just putting in a prompt and taking the finished product and being done with it.
Image Source: Westworld
The issue finally made headlines in August of 2022 when Jason Allen submitted an A.I. piece titled “Théâtre D’opére Spatial” to the Colorado State Fair and won first prize. Technically, the category he competed in was a new one that they had set up called “digitally manipulated photography” and he had let them know up front that he was using an A.I. However, many people felt that he was not upfront about how much of the A.I. he was using. Allen paid for Midjourney and spent 80 hours making more than 900 iterations of art, putting in so many keywords and prompts, and playing around with it to get what he wanted. He then had to crop, digitally paint, and add details (like a head that was missing!) using Gigapixel A.I. as opposed to doing it all by hand. The result was so stunning that a lot of people (including the judges) did not realize it was A.I. (at first). Many of the other contestants felt cheated that artwork that they had fully created digitally without the use of an A.I. did not win, although a few did still say it was impressive. Did Allen technically cheat his way into winning the competition by using A.I.? Some might say yes since their artistic visions and hours of work are all down the drain now. On the flip side, looking at the hours of work that Allen did do, he did not just put in one prompt and have it spit out the perfect image. In the end, the main issue does remain. It was an A.I. that ultimately created something using prompts fed to it.
One of the other fears that people are having with this development is the use of A.I. for school projects or published works. If you put in a writing prompt for a paper or a novel into ChatGPT, fix up some grammar, and zhuzh up sentences a little, will it be easy to figure out it was an A.I. who wrote it and not you? How will that affect teachers and book publishers? What about Hollywood scriptwriters? Not to mention art or writing contests. There is nothing to say that you cannot have A.I. generates something and put your name on it as your art. This is what worries a lot of artists who have worked years on projects only to be confronted with something that can easily steal their life’s work for another to put their name on.
With every new technological advance, there is discussion among the artistic community about their creativity becoming threatened and their jobs becoming obsolete. Many traditional animators felt this way with digital animation. Many digital animators felt this with the rise of CGI. Many traditional novelists felt this way with the rise of self-published books doing better than their own. Through it all, artists have persevered. People still buy art. People still support independent artists and writers, and companies still need them. The rise of A.I. is still worrisome with how it has developed so quickly, and continues to do so, but that should not stop artists from making their own work and continuing to follow their passions.
Image Source: CultureSlate and Anthony C. using Stable Diffusion (left) and original ‘Ten Lords a Leaping’ (right)
There is one silver lining. Technology can never actually put its own creativity, its own je ne sais quoi into the work it pumps out. Even if (when) A.I. grows even more, it will not be able to capture what you as an artist can capture. When you create art, you pour your heart and soul, your dreams, your fears, your laughter, and your tears into your art. No artificial intelligence can do that. Even if a painting is pretty, even if a story seems well written, there will be no feeling behind it, no meaning, no story behind the piece of art. It will just be there.
The world may try to hurt artists or try to pretend that it does not need art anymore now that A.I. can create it as well as it does. But they are wrong. Artists are still needed in the world. People will still crave real art, not the stuff made by technology. Artists should keep on making art. People will still want your art. A.I. cannot replace the genius and uniqueness of each human brain.
A.I. is changing the landscape, there is no denying that. Will it be for the better or for the worse? Who knows. Time will tell. There are good and bad sides to A.I. art technology, and it will be interesting to see how it changes and evolves over the coming years.
READ NEXT:
Latest News
Featured
