Using advanced algorithms, Argentine artist Sophia Crespo invents the insect, while American Robbie Barat distorts the nudity of classical art. Drawing with artificial intelligence will disrupt the visual creation, according to its defenders. “It’s like a ballet between man and machine,” said Jason Bailey, collector and one of the most famous crypto-art bloggers. For the most part, these digital artists work with programs known as supercomputers and generative adversarial networks. These are two artificial intelligence algorithms, called “neural networks”, which compete to provide the artist with the most complete image: man first provides the source images and adjusts the parameters to obtain from them, a result that appeals to him. Sophia Crespo, 30, uses it to recreate animals. The aim is “not to avoid true nature, but to connect with nature through a medium where we spend a lot of time, which is a digital medium,” he pointed out in a video interview from Lisbon. Its insects are ultra-realistic, with antennae, wings and bodies that look like something outside of an entomology textbook. All but one of them are missing a head, and their bodies seem to have gone through multiple genetic mutations.
– All artists?
The astonishing advances in AI suggest a world where computers would be able to learn and create like humans. But, for the moment, artificial intelligence still needs guidance and countless trajectories between the artist and the proposed models of the neural network for the Sophia Crespo series of insects. If “computers are an integral part of the creative process”, “the ability to create realistic images does not make everyone an artist”, a quality that requires “a critical and innovative ability”, according to Pompidou’s curator Camille Langlois. Robbie Barrett began his career in 2018, as Sophia Crespo. He enters thousands of classic art photos into his computer, and starts a conversation with the machine, until he finds what he was looking for: a series of shapeless busts, half way between Salvador Dali and Francis Bacon. “When I work this way, I’m not creating an image. I’m creating a system that can reproduce images. In a way, I’m creating a tool,” he says. At age 22, this developer is a digital art prodigy. One of his works sold for more than 700,000 euros at Sotheby’s in March. Four years ago, his code was widely used by the French collective Obvius to create a “portrait of Edmund de Bellamy” painting, which at this time sold for more than 400,000 euros in Christie’s.
– Lawyer’s chair
But today, simple new algorithms called “Transformers” are preparing to wreak havoc on this new universe. One problem so far has been “placing text in input and output images”, Hugo Castles-Dupre and Gauthier Vernier explain from the clear sum. In particular, the instrument needs to be able to take astronomical quantities of all kinds of images with descriptions. A Titanic task is only accessible through well-funded projects, such as the California start-up OpenAI to the Dall-E 2 model, specifically billionaire Elon Musk, or Image, a competing project funded by Google Research. From a simple sentence, the machine is then able to mix ideas and create several representations of a “couple of mullahs who skateboard”, a “lawyer’s chair” or a “monkey astronaut”, all in the style of a photographic, comic book. Or in the style of a 17th-century Flemish painter. “It’s generally the best in terms of image generation,” said Sophia Crespo, who was able to experience Dal-e. Experts believe that these programs could revolutionize the entire image creation and editing industry. Many examples have already spread on social networks, including the annoying presentation of an animal at the junction of shrimp and centaur from a laboratory called Midjarni. On the other hand, OpenAI and Google have not yet released or marketed any consumer tools directly, especially due to the risk of malicious use.