how AI artists want to revolutionize visual creation

2024-02-22 09:46:25

These “AI artists” rely on generative artificial intelligence software in their creative approaches. And this despite very strong criticism from cultural circles against this technology, accused of undermining copyright.

“With artificial intelligence, we can develop our imagination as much as we want”assures photographer Polina Kostanda alias Polly, before specifying: “Now I can’t live without it”. The young woman even defines herself as an “AI artist”: to design her visual creations, for the past year, her main tool has no longer been her camera but the software equipped with artificial intelligence, Midjourney.

This allows any user to quickly generate images or fake shots, based on the written commands they enter on its interface. Similar tools have multiplied in recent months, with software like DALL-E or Stability AI. All are trained on massive databases of images to make them accurate in the visuals they produce. More recently, for example, it was the test version of the Sora software from the company OpenAI (at the origin of ChatGPT and DALL-E), capable of generating videos from a simple textual query, which panicked the Cloth.

On her Instagram account, artist Polina Kostanda aka Polly, creates a wacky universe based on her work on the Midjourney software. Screenshot Le Figaro

With these images of a new genre Polly wants, on her own scale, to titillate her audience. “Through humor or provocation in my images, I encourage people to think about current issues”, she describes. His Instagram account mixes burlesque photos of insect creatures and deformed faces.

To achieve this result, the artist writes descriptions (also called “prompts”) which she inserts into the Midjourney software and the latter materializes her idea almost instantly. “I often rework my texts for a whole day, so that the result really corresponds to my vision”she explains, “but it’s worth it: the process is truly fascinating”.

A fledgling business

Artistic agencies, like the very young desyre.AI, even see it as a new business. Launched at the beginning of January, the structure in question is made up of a team of«AI artists» coming from Berlin, Madrid, Montreal and even Barcelona. Objective, to offer artistic services based on artificial intelligence to clients from “photography, from the cosmetics or fashion sector”explains founder Sophie Hiller, who specifies that the initial budget for a project is estimated at 500 euros. “The price is calculated based on the number of images requested, usage rights and the size of the company”she explains.

The Parisian artistic agency AKQA does the same. “When clients ask us for a visual, I create them using Midjourney before working on them again and I sometimes use ChatGPT to present them with the storytelling around the project”, explains artist Fan Shi, who works for this agency. According to the young woman, certain original publications generated by this software can be worth up to « 900 000 euros». “For me, we have entered a new era, where AI is now an integral part of the creative process”insists the one who, on her Instagram account, also offers creations imagined using the same tools.

Revolt of artists to protect their works

“A strong creative vision is necessary for this use of AI in art to continue over time”however, nuance Léa Saint-Raymond, historian and head of the art market course at the Louvre school. “These tools allow anyone to produce visual creation quickly. Some only use it as a gadget and call themselves artists.she insists. “A bit like at the time when photography appeared, when everyone could claim to be a photographer”.

On X (formerly Twitter), many artists share their dismay over the success of this software. Above all, they warn of the looting of their work. Because the immense databases from which these AIs learned to compose images are made up of works designed by humans, and sometimes protected by copyright. “Someone trained an AI with my works without my consent”testifies for example the designer SimzArts on her account. “If you love me as an artist, shouldn’t you respect my work?”she adds.

Three American illustrators decided to file a complaint a year ago against the groups Midjourney, Stability AI and Deviant Art. As part of this procedure, they filed a document which allegedly belonged to Midjourney. This contains 16,000 names of artists and works which would have been used to train the software’s artificial intelligence model. There we find the names of famous illustrators like Hergé or Sempé but also painters such as Man Ray and Frida Khalo. Also in Europe, a collective of artists, the European Guild for Artificial Intelligence Regulation (Egair), was formed to assert its rights against these American companies.

The revolt does not stop there. In January, researchers at the University of Chicago launched a tool ititled “Nightshade”. CThis allows artificial digital data to be added to artistic works, which deceive the perception of artificial intelligence. So, for an image of a cow in a field, “an AI model will perceive a huge leather bag lying in the grass”, relates the researchers behind the project. This “data poisoning” makes it possible to interfere in the development of artificial intelligence and its ability to reproduce images.

Going beyond simple image reproduction

So, agree to make your job easier or ban this software forever? If the question is not resolved, some want to believe in the possibility of creating real creative processes from artificial intelligence. This is the case of artist Justine Emard, who creates her own AI rather than using Midjourney or DALL-E. “I train my models on databases that I completely control. I want to be at the origin of what will animate my work afterwards, she emphasizes.

From this process, the Frenchwoman created a work entitled “Supra Organism”. “To make this work possible, I designed software that imitates a swarm of bees and learns from observing a real swarm. Little by little, he is able to make his own predictions.she describes. “My software is trained with data that I went to find myself,” adds the artist. “This learning process that I worked on is entirely part of my artistic approach”.

To promote her work, Justine Emard must do teaching with certain cultural institutions. For example, she spoke at the Parliament of Photography at the Palais de Tokyo in Paris, to talk about images and artificial intelligence. “We are entering a new stage of creation, where we must rethink the notions of copyright and respect the consent of artists”she explains. “This is a delicate issue and we are currently in the heart of this debate”.

Curiosity of museums

But the public’s curiosity is already enough to popularize some of these works. At least, that’s what the organizers of the first Biennale dedicated to art produced with artificial intelligence say, which will take place in several cities such as Warsaw and Berlin in 2024, then New York the following year.

“We have just launched our website and our Instagram page and the public reaction has been beyond our expectations”testify the latter, who all specify that they evolve in the artistic and technological sphere without revealing their identity. “People are intrigued and captivated: it’s as if we have opened the door to a new dimension of events”.

These new installations even end up being acquired by renowned galleries and museums. THE “Supra Organism” by Justine Emard was sold to international collections in Germany and South Korea. The Museum of Modern Art (MOMA) based in New York also acquired a work by Refik Anadol. The Turkish artist is known internationally for integrating data generated by artificial intelligence into huge digital canvases, which represent “dreams and hallucinations” of the tool used.

1708596370
#artists #revolutionize #visual #creation

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.