Artificial Intelligence & painting, a paradigm shift?

Artificial Intelligence places us, in my opinion, in a new paradigm shift in the field of artistic creation. For the first time in history, artistic creation is not something exclusively human. These lines are intended to understand where we are, and where we are going.

Artificial inteligence generations based on painting.

The precedent: iconosphere of the Internet and the social media

The last big paradigm shift in the visual representation field was, in my opinion, at the beginning of the current millennium, when compact digital photo cameras became popular. These cameras led to angles, frames and flash lighting that until then were not so frequent. The amount of images created was multiplied exponentially, while its importance was reduced. In the world of analog photography one thought twice before shooting.

In the year 2004, I was a second year art student at the University of Granada, Spain. That year, a young Juan Francisco Casas won the ABC painting price, thanks to a painting that incorporated all the characteristics of digital photography. At the time it was something new, I even dare to say that was something revolutionary. Fascinated, I changed my way of painting almost immediately (see artworks created while I was still studing).

A few years after the popularization of digital cameras, the use of Internet and social media was generalized. This was the germ of the current iconosphere, characterized by saturation of decentralized content and the concept of viral. The image is cheaper and at the same time more important than ever. Today, there are many painters who incorporate characteristics of images taken by digital cameras or smartphones into their artistic language, such as myself. In 2005, painting as a digital photo was something bold. Today, it is a widespread practice.

Puri y Juanma Moreno
Example of painting with obvious influence of digital photography. See the angle, the frontal lighting of the flash and the shadows it generates, as well as the freshness and spontaneity.

Artificial Intelligence as a creation tool

Neural networks have the ability to understand an image and decompose it semantically, like a human brain would do. They also have the ability to find relationships between concepts and represent ideas based on their experience. They can create new images, and, interestingly, generate their own shapes with unique characteristics. That is why I understand that there is potential for content generated by neural networks to influence the content generated by human artists from now (2019) onwards.

That is why I believe that, just as digital photography changed everything in 2005, Artificial Intelligence will do the same in 2020.

Artificial inteligence based on painting. Juanma Moreno Sánchez
Artificial inteligence generations based on painting.

I am currently conducting some experiments “dialoguing” with neural networks capable of understanding and creating images through painting. The conclusion of these experiments is that the process of painting is very similar if not identical in the head of a painter and in an adversial generative neural network (GAN). Creativity is, in essence, creating new connections between neurons. And both natural neurons and nodes of artificial neural networks have that ability. Technology imitates nature (see perceptron).

Artificial Intelligence and the monstrosity

When a neural network generates an image of a cat or a car, this representation can pass through a real photograph. When it comes to a human face it is different. In that case, any deviation of half a centimeter from a line ends up giving a monstrous appearance. But these monstrosities are not perceived in the case that the photos are, for example, of cats. The explanation is that our brain is super-specialized in human faces, and not that specialized in cat faces. This means that there is a path for the improvement of neural networks towards more human parameters, but in the meantime, let’s enjoy the beauty of the monstrosities it gives us.

Monstrosities generated by an artificial intelligence based on the painting shown below.
Reina, Juanma Moreno Sánchez
Queen, Juanma Moreno Sánchez. Painting from which artificial intelligence has generated monstrosities

Seeing some monstrous generations of neural networks, it’s easy to remember the work of artists like Francis Bacon. I believe that there will be a renewed interest in the monstrous, the deformed, and the world between abstraction and figuration. The day will come when the improvement of neural networks will be able to create hyperrealistic images too, opening, again, new expressive paths to be determined..

I understand these technologies not as substitutes, but as helpers of the artist. It is in our hands to be able to squeeze their potential, and that languages ​​evolve into new forms consistent with our time.