Will artificial intelligences replace artists?

This is a scenario that could become commonplace in the coming months. Luis Martell, a Mexican illustrator, explained in early January On twitter being hired by a couple to illustrate a children’s book. After receiving a down payment of 50% of the price, the couple announces by e-mail that due to “budget cuts”, the drawing that Luis was going to be paid for will finally be done… by an artificial intelligence. Free of charge.

2022 has been a year of mind-blowing progress for artificial intelligence. Today, Anyone can create professional looking designs in seconds. and no special abilities. Enter a few words in an algorithm like Dall-E, Midjourney, or Stable Diffusion to describe the desired scene or style. Some of these creations even win art contests and they are exhibited in galleries.

But as the quality of the images created increases, concern is growing among some professional artists. What will become of the human creators? How to live with systems capable of generating very complex works in a few seconds, without the need for technical knowledge and for a ridiculous cost per image?

For Karla Ortiz, concept artist who has notably worked for Marvel or Ubisoft, this scenario is not a distant danger. “AI (by artificial intelligence, editor’s note) has already begun to replace artists for certain works,” he laments to Tech&Co, who was notably involved in imagining the appearance of strange doctor in the Marvel movie.

She cites, for example, the San Francisco Ballet, which used the Midjourney tool to create promotional campaign images for its performance of Nutcracker. “If we lose the ad market, it’s one of our biggest sources of revenue that flies away,” explains Deb Lee, an artist who campaigns against the current development of these algorithms. “And we don’t get paid enough anymore.”

The use of AI that create images is gradually progressing. She wins magazine covers…

…from books…

…article illustrations…

… but also animation or videogames.

“I have friends at a video game company who were forced to use AI, and when they challenged that decision on ethical grounds, they were told they were wrong because everyone didn’t care if the original images were copyrighted or not, and that was I was going to use no matter what,” says the artist. He created the hashtag #CreateDontScrape (“Create, Don’t Scrape” *) to bring together artists who oppose the current development of image creation algorithms.

Cartoonist Victor Dulon, who has worked with Marvel Studios and Lightning, since this year he has been introducing his students to Midjourney and Stable Diffusion in the art schools where he teaches. According to him, not all creative professions will be affected in the same way by the development of AI.

“AI allows you to speed up the concept creation part, generating a lot of images quickly. You can create around fifty logo examples in just a few minutes,” the artist explains to Tech&Co.

The more mechanical tasks, such as the work ofinterventionist, the animation helper that draws intermediate frames between the different “keyframes” of the cartoon. Or those based on the mastery of a particular skill… Like a specific art style, which the AI ​​can train to reproduce, for infinitely less than its creator.

Deb Lee knows it. The artist discovered that an AI available on Reddit had been specifically trained to create images extremely similar to his drawing.

when the artist he was moved by it on social networks, the show’s creator deleted his post. But this custom AI has since been republished elsewhere, Deb Lee tells Tech & Co, without the artist being able to keep track of this AI coming into direct competition with his work.

Deb Lee’s case is far from unique. Internet users have created a multitude of models to imitate the work of a famous designer, Sam Yang, to such an extent that a competition has been organized between these algorithms to decide between them.

And that’s not to mention Greg Rutkowski, a Polish artist whose name has been used tens of thousands of times by netizens to create images in his style, according to the MIT Technology Review.

“We use our work to put ourselves out of work,” laments Karla Ortiz. The Conceptual Art Association, where she sits, launched a crowdfunding campaign to advocate with US legislators and review the legislation surrounding these imaging programs.

By generating an infinite number of more or less random images, AIs can, for example, stimulate the inspiration of human artists. “Dall-E can help to see things in a different way, present interpretations and details that he may never have thought of,” OpenAI describes.

“You can also train the model on your own creations, and thus generate works that correspond to your style,” explains Charlesai to Tech&Co, one of these new “AI artists” that use image creation algorithms. “It allows you to generate composition ideas, colors…”

And using AI doesn’t mean relying entirely on them. “To get the most out of these tools, you have to be an artist, understand things like composition, style, or proportions,” says a spokesperson for Stability AI, the company behind the Stable Diffusion software.

Some of the most talked about AI images, like the painting that won an art contest (cover photo), were extensively reworked by a human artist to arrive at the final result. Images from the graphic novel Zarya of the Dawn, the first AI-created comic to be copyrighted in the United States, they were also organized and written by their human creator.

Will AI create new professions?

If jobs are going to disappear, for AI supporters, it’s just the story of technical progress. “We know that AI will have an impact on jobs, just like the computer before it,” the OpenAI spokesperson told Tech&Co. “The nature of many jobs will change, new ones will be created and others may be eliminated.”

Victor Dulon compares the arrival of these algorithms with that of Ikea in the field of furniture. Production that can be automated and standardized will be, and the remaining human artists will focus on fancy crafts.

Some positions could even come out stronger from the development of AI.

For some, AI could even create new professions. Among them, “punctual engineering”: that is, knowing what words to use to propose the sentence to the AI ​​(called “prompt“) perfectly corresponding to what we want to generate, among the infinity of usable words. And thus make the most of the potential of these AI.

But this knowledge is highly dependent on the algorithm used, because they are all programmed differently and therefore interpret the prompts differently. And not everyone agrees on the value of this knowledge in a professional context.

Most of the artists interviewed agree that image creation algorithms will not make the artist role disappear. “An artist is someone who shares her vision with the world, whatever the tool,” says Charlesai. But the question posed by opponents of him is: “who will be able to continue living on him?”

Generating an infinity of human-quality images in the blink of an eye for an extremely low unit cost, everything seems set for imaging algorithms to overwhelm the output of some of today’s human artists, their competition, and therefore reduce your income

An unthinkable situation just a few months ago. “We thought that artificial intelligence would save art, that creativity was something specifically human,” laughs Victor Dulon. “But AIs like Midjourney are already creating crazy poetic stuff. We’re going to have to scrap this vision of the artist. We don’t have a choice anymore.”

* “Scraping” is the technical process by which the billions of pieces of content found on the Internet are collected and used to train creation algorithms.