Art & Artificial Inteligence, a two-way street (first steps).

The idea is to implement a new experimental creative process, that seeks to exploit the possibilities of Artificial Inteligence to lead the way in the artistic creation field.

On this new methodology, the starting point is artist’s recent artworks. Based on that, a set of neural networks, with different capacities, will produce new content, that is also self-referential. This result will be as well reinterpreted by the human artist. This two-way process encourages the debate about the artistic language, about the machine generated visual language and the interaction between them, in one and the other way.

The results of this methodology can be as well the starting point of another iteration of the loop. That opens the possibility to a virtuous circle of creativity.

Creaciones humanas sirven de punto de partida para la red neuronal, que a su vez sirve de inspiración para otra creación humana, y así sucesivamente.
Human creations are the input of the neural network. The result of it is, at the same time, the starting point of another human creation.

This is an experimental project, focusing on processes, whilst not forgetting the crafty part or the artwork as an object.

Secondarily, this proyect has the potential to normalize the use of Artificial Intelligence on the artistic field, something that did not happend yet, I think because of the lack of knowledge.


The starting point is the photographic documentation of my previous artworks. One of the paintings will be the origin of the generated images.

Painting selected for the first loop of the methodology
Starting point, Juanma Moreno Sánchez, 2019.

A open source GAN (Generative Adversial Network) will process the images, extracting semantic features, and generating new “child” images taking those features into consideration. Child images are completly original pictures, but they look somehow familiar.

Different variants created by the neural network, based on the real painting.
Results of the neural network, given the original painting.

One ot those results, created by the Artifical Inteligence, is going to be reinterpreted again, this time by a human inteligence, painting it manually. Fundamentally, what a human painter does is similar to what a neural network does; select features and recreate them throw a custom filter, that in my case is the technical ability. The painting created by the human is “child” of the image created by the neural network, and “grandchild” of the first painting.

Painting by Juanma Moreno Sánchez based on the images created by the neural network.
Painting based on the images created by the neural network.

This steps can be repeated multiple times, and create whole genealogies of images, altering human and artificial inteligence.

Variantes creadas por la red neuronal basadas en la pintura que a su vez está basada en las creaciones de la red neuronal basadas en la primera pintura.
Variations created by the neural network based on the painting that is, at the time, based on the neural network results, based on the first painting