Why did you do the exhibition
A neural network is just an algorithm written by people to accomplish certain goals and tasks, but people who are far from IT do not understand this very well.
There is a lot of hype around the term "Artificial Intelligence" now. The media add fuel to the fire with headlines like "AI is the future of everything", "AI will change everything" or "Artificial intelligence will take over the world." With this project, we wanted, among other things, to raise awareness among people. Emphasize that artificial intelligence is not a threat or magic that produces an absolutely random result or, on the contrary, the only correct solution. In the texts about the exhibition, we definitely say that we use not just “artificial intelligence”, but a generative adversarial network (GAN for short) and, whenever possible, try to explain the general principles of its work.
Modern man is often faced with the applications of machine learning. Take the same social networks that study our behavior and show what they think we would like to see. It looks harmless, but it gives rise to real social problems. More - for example, in this documentary: The Social Dilemma (Netflix, 2020).
We believe that such things need to be conceptualized, discussed and jointly developed new forms of digital ethics. And everyone needs to be discussed! Nothing will work as long as a wide audience imagines a threatening black box in which something is happening , and which will soon replace us all. Art is one way to show what this mysterious artificial intelligence is.
What were inspired
Of course, we were not the first to think about using the GAN. In 2018, the "Portrait of Edmond Balamy" created by a neural network was sold at Christie's for $ 432,500 .
"Portrait of Edmond Balamy" appeared in the right place at the right time. It was bought at a price that was 45 times higher than the experts' estimate. It was the first such precedent, a chance to enter art history, but it was not the beginning of a trend. At the time of this writing, human artists are still outselling, and this is unlikely to change soon.
There are other projects, there are many of them. Personally, we love Trevor Paglen's work the most . All this became interested in the St. Petersburg artist Ivan Plyushch. Together, we picked up datasets of pictures on different topics and ran them through StyleGAN2 . Ivan looked at what was obtained, used these results as sketches and created paintings based on them.
Now in more detail.
How the works were created
This is what we used to find and download thematic images that formed the basis of the dataset. Then, with the help of a little magic, the images were adjusted to the desired size and format and sent to the neural network for training. Each dataset consisted of approximately 600 photographs.
It turned out something like this:
StyleGAN2 and, more recently, StyleGAN2-ada were used to create works from this series : generative adversarial networks developed by nVidia and giving some of the best results to date. The thing is not new. Those who follow the development of neural networks or media art have probably stumbled upon at least the https://thispersondoesnotexist.com/ project , which works on the basis of these grids.
Ivan Plyushch looked at the results and painted the following on canvas:
Oil and acrylic on canvas. 159 x 129 cm.
Oil on canvas, acrylic. 180 x 180 cm
Oil and acrylic on canvas. 85 x 85 cm
In total, we made several episodes: "After the Chair" (this includes the three previous pictures), "After Me", "After You", "After the War", "After the Weapon", "After Religion", "After the Politics" and other.
Here is a picture from the "After Porn" series:
Canvas, acrylic, oil. 180 x 180 cm
And this is how Ivan drew before he got acquainted with the neural network. He had complex, viscous, flowing into each other forms before, but different:
How the premises were prepared
The contacts of the curator and the fact that Ivan Plyushch is a fairly well-known artist helped us to find the premises (his works are now shown, for example, in the Russian Museum at the exhibition " Generation of Thirty "). It was agreed to hold the exhibition at the Sergei Kuryokhin Center for Contemporary Art. It is a reputable organization in the art scene.
Six rooms with a total area of about 400 square meters were provided for the exhibition. There we show 25 works by Ivan and separately - the results of StyleGAN2's work. Some of the images created by the neural network were printed and hung at the exposition, some were collected in a video that is projected onto the walls.
The full title of the exhibition is “After all / After all. Part 1. After the chair ”. A chair is something clear, familiar and as simple as possible. Using this example, it was convenient to explain how we worked on the project. And this theme also refers to the work of Joseph Kossuth, symbolic for modern art, who showed at the exhibition a real chair, a photograph of this chair and the definition of the word “chair” from the dictionary. This example is usually used to explain what conceptualism is. You can read it here .
We plan to show the project after St. Petersburg in Moscow and in Nizhny Novgorod: these will be the next parts of the project.
Downstairs, on the floor, the curator and the artist decided to place sawdust and pieces of microcircuits: a metaphor for the work of a program that “split” a chair, “reworked” and returned it to reality.
There is such a psychiatric term - "apophenia". Apophenia is the ability to see structure and relationships in random or meaningless data. We feel apophenia when we look at a cloud and see that it looks like a horseman. Neural networks interested us in the fact that the result they offer makes us constantly feel a feeling of apophenia, and with it - a slight anxiety. It was fun to work with. The sketches were generated by the program, and the finished works were created by the artist: in this way, the empirical experience of a person was added to the work of the neural network.
It is unlikely that a visitor will leave the exhibition with a sense of harmony and tranquility that classical art can give. Most likely, the questions will arise "How to understand and interpret this?" Just like with neural networks. If the viewer takes away curiosity, slight anxiety, and ideally google about GAN, we will assume that we have managed to implement what we have in mind.
The exhibition runs at the Sergei Kuryokhin Center for Contemporary Art (St. Petersburg, Ligovsky pr., 73, 4th floor) until April 15. Come!
Artist: Ivan Plyushch
Exhibition curator: Marina Alvitr
Producer: Ivan Puzyrev
Technical support: Artem Konevskikh
Helped by: Anna Padalyako, Daria Elizarova, Evgeny Novikov
Wrote the article: Daria Yakovenko