By Gary Symons
TLL Editor in Chief
Artificial Intelligence apps like Lensa or ChatGPT offer great promise, but they are also a threat to creators.
That’s the very real fear sweeping through the artistic community these days, as writers and illustrators cope with AI apps that can create in seconds what takes them hours to do.
The technology also poses serious challenges for the licensing industry as a whole, but right now, it’s the creators of visual arts who are most at risk.
Consider the case of the popular Korean artist known as @ato1004fd. The artist livestreamed an 11-hour sketching session so their 22,000 followers could watch the process of creation in real time as they created an image of a character from the video game Genshin Impact.
By the time the piece was completed, a viewer had screen-grabbed an image, and used an AI app to finish it, literally in seconds. He or she then posted that version to social media, and accused @ato1004fd of being the copycat.
The story, which I found on Artnet, is the first I’ve seen of people using AI to beat illustrators at their own game, and the implications are disturbing. While AI may not come up with the ideas for art, it can turn out artistic works with such blinding speed that no human could possibly compete.
As Liz DiFiore, president of the Graphic Artists Guild, told Artnet, “It’s the Wild West at the moment. People are primarily worried about their jobs.
The Graphic Artists Guild says the median salary for an illustrator is $50,000 USD, but wages have been decreasing for years, which DiFiore blames on copyright infringement. “Wages have been decreasing over the years,” she explained. “And infringement is probably the biggest reason why illustrators are seeing the value of their work fall.”
The Artnet article is just one of many that have arisen in the weeks since the AI app Lensa was released on mobile devices last year, and it’s only one of many. The art imposter described above used another service, called NovelAI, which has looser rules around copyright infringement. Essentially, it offers an “image to image” service which allows you to upload one image, and use text prompts to have it alter the image to your wishes. For example, you could alter the Mona Lisa to wear modern clothing, or the Botticelli’s Venus to hold a can of Budweiser.
Curious about the implications of this new technology, I used the online app Dall-E to create a number of illustrations, such as a robotic reporter writing on a laptop, illustrating my own obvious fear that an AI app could replace journalists like me. I also created one of Mickey Mouse dancing with Spongebob Squarepants, which took all of 10 seconds. That alone made me wonder, how do companies protect their brands when literally everyone in the world can create images in seconds, and post them anywhere? Do you just sue … everyone?
The other issue is that these apps use existing art as the means to train their AI applications to create artistic works in various styles. That brings us to the other side of the AI art coin, which is the allegation by many artists that their work and signature style is being copied by tech companies, without any compensation.
A prime example is the famous fantasy art illustrator Greg Rutkowski, who discovered last year that his name was one of the most popular prompts on the AI art platform Stable Diffusion. The illustration at the top of this article was created by Rutkowski for Hasbro’s Wizards of the Coast division, which is yet another illustration (no pun intended) of the potential licensing implications of AI artbots.
On the bright side, one could consider it a compliment when you’re more popular as an art search term than Picasso, but not when literally any untalented slob like me can create art in your style in mere seconds, with little effort.
“The only thing that could at least stop feeding the algorithm is to stop posting your work on the internet, which is impossible in our industry,” Rutkowski points out.
The problem is, there’s very little protection for the style of artists, as opposed to doing an exact copy of a particular copyrighted work. In another test of AI proficiency, I asked ChatGPT to create a press release about two companies in the style of Dr. Seuss. The result was, I hate to say, scarily good.
Megan Noh, an attorney and co-chair of Pryor Cashman’s art law group, says there are few precedents supporting the cause of artists when it comes to copyright infringement and artificial intelligence.
“The field is very dynamic and evolving,” she said. “An artist does not have copyright protection over style,”
From a licensing perspective, AI brings up far more widespread concerns. Right now the main concerns are being brought up by visual artists, but in fact, AI has been used to write short stories and novels, it’s regularly used to write news stories (although not by TLL, just sayin’), and it’s even created short films.
When I was in university, my writing prof once said, “Art is how human beings tell their own stories, to themselves.” The question facing us as an industry is whether we want our own stories told to us by machines? And if so, how do the slow, cumbersome human creators protect their place in the world of art, film and writing?
From a licensing industry perspective, the same question could lead to an AI arms race. As AI improves and speeds the creation of artistic works, how can a company possibly NOT adopt AI and still compete head to head. And just as seriously, if literally anyone can create derivative works with a few keystrokes, how do you stop a tsunami of copyright infringement?
Today there are few rules that create boundaries for the use of AI in this arena. These early example of the power of AI, however, show that the arts and licensing industries will quickly need to make decisions on what is and is not allowed, and fight for the ability for human artists and creators to survive and prosper.
There certainly are many legitimate and non-harmful uses for AI in the creator space, of course. One journalist, for example, wrote about using ChatGPT to quickly create a list of interview questions on a complicated topic, saving at least an hour of work and research.
But that’s very different than having AI bots replace human storytellers altogether, and that is the decision we are faced with: Do we want to tell our own stories? Or do we want to be passive receivers of stories told by server-based AI bots?