Canvases created with artificial intelligence exhibited in Paris: “It’s a dialogue with the algorithm”

The Obvious collective is exhibiting a series of paintings inspired by the Seven Wonders of the World until January 14 at the Danysz gallery, in the 11th arrondissement of Paris. Artworks created using AI algorithms.

This is one of the first times that paintings created with AI have been exhibited in an international gallery. 7.1the new series from the artist trio Obvious presented at the Danysz gallery, includes 14 works in total. There are first seven paintings for each of the Seven Wonders of the World: the pyramid of Cheops at Giza, the hanging gardens of Babylon, the statue of Zeus at Olympia, the temple of Artemis at Ephesus, the mausoleum of Halicarnassus, the Colossus of Rhodes and the Lighthouse of Alexandria.

We worked with a historian to try to bring together all the architectural descriptions of the Seven Wonders of the World in the ancient texts, to create our own interpretation of the monuments. Then from a document of a hundred pages, we built the scripts for the algorithms, says Gauthier Vernier, member of the Obvious collective.

The artists then created their “quick“: the sequences of words used by the AI ​​to generate images. The collective thus worked with modified versions of algorithms accessible to all such as DALL-E and Midjourney, by adding their own parameters in the code. These so-called programs “text-to-image”, whose popularity has exploded in recent months on the internet, make it possible to transform text captions into visuals, by “training” from a huge online database.

A new man-machine language is being developedexplains Gauthier Vernier. The simplest example is the Google query. When you do a search, you don’t ask Google the question as if you were talking to a human being. We will organize the sentence to get the best possible response from the algorithm.

When you create with the new text-to-image algorithms, it’s a bit the same principle but much more precise, he continues. We are going to look for certain aesthetics and certain formats. And we will be able to forge words, for example by trying to make an algorithm understand the notion of ‘colorful’ in order to request specific results.

The collective then worked with a workshop of copyist painters “to go from the digital format, from the image obtained on a computer, to the physical format with copies in oil on canvas“.

The complex project was launched a year and a half ago. “The technology of these algorithms has evolved so quickly that we had to start the project from scratch three times. There, we have reached a point where we are satisfied with the result, and we say to ourselves that seeking more realism would not necessarily be more interesting. We like to keep it a bit trippy, we can see that the aesthetic was created using algos. For example on the statue of Zeus, the face is a little erased“, indicates Gauthier Vernier.

The process is labor intensive, he adds. It’s called prompt-engineering, the art and technique of creating ‘prompts’ to optimize the use of these algorithms and make it understand what you really want to create. We are always a little surprised, it is difficult to predict what will come out of the algorithm. There is a part of randomness and outsourcing of the creation, keeping control over the message and the aesthetics. It is a dialogue with the algorithm.

The member of the Obvious collective highlights the multiple artistic choices that played on their representations of the Seven Wonders of the World: “Typically for the pyramids, texts evoke the reflections of light on the walls – with a very clear and smooth stone which covered what we see today. We decided to emphasize this phenomenon, rather than sticking to a ‘real’ geometric construction.

Gauthier Vernier also responds to the criticisms aroused by the development of these algorithms: “With the appearance of these algorithms, the question arises of the data used to train them. Potentially any work of art published on the internet can be affected. We ourselves cannot know which images were used to train the algorithms that generate our visuals. But if no data is recognizable in the final result, it can be considered a work of art in its own right, copyrights are respected. The idea is precisely to play with these algorithms to create a new, transformative aesthetic, it’s experimentation.

When we started, a lot of people told us: ‘Robots are going to replace artists’, ‘It’s like Terminator’… We know that part of our job is also to explain how all this really works, to demonize AI by teaching“, says the artist.

Arguments supported by Magda Danysz, who founded the gallery in 1991: “Some people are skeptical of art with AI, fearing that the machine will replace humans, although it is a very interesting artistic field with a potential market. With the success of smartphones, some predicted the death of traditional photography. As with other areas, AI is a complement, which can help inspire talented artists who know how to use the tools of the moment. The role of our gallery is to accompany these artists by giving them the means to meet their public. Our mission is to lead people towards new forms of extremely contemporary creations.

Beyond the canvases, the Obvious collective’s series also includes seven digital works designed as extensions of the Seven Wonders of the World, sold in the form of NFT. “These are digital works that can be discovered by downloading an augmented reality app and scanning the paintings. The idea is to make the link between traditional art and digital art. We created these works with algorithms that allow for outpainting. We start from an image and we can extend our environment by creating visuals around it, square by square, describing with text what we want to see appear“, indicates Gauthier Vernier.

A pioneer in artistic creation using AI, the Obvious collective has notably sold in 2018 at Christie’s a portrait created with algorithms that analyzed thousands of works, for nearly half a million dollars. The French also created other series centered on Japanese prints and African masks. The next project of the collective: launch a research lab to imagine new creativity tools with AI.

We would like to thank the author of this write-up for this awesome web content

Canvases created with artificial intelligence exhibited in Paris: “It’s a dialogue with the algorithm”

You can view our social media profiles here as well as additional related pages here.