Artificial Intelligence: art, artist or medium? 1

Generative artificial intelligence puts us in front of a debate that shakes the entire creative sector
It is possible to say that art has always resorted to technology to materialize itself in the world. An artist's expression may be possible through a brush, a gouge, an organic material or a computer.

However, the relationship between art and technology leads to debates about its limits in each new language, just as photography shook the artistic scene in the 19th century, the arrival of generative Artificial Intelligences, such as ChatGPT, Midjourney and Dall-
And moves the controversies of the present day.

These are tools that generate images from texts or other images and place us in front of a moment in the history of art in which technology does not appear only as an extension of the body but acts as the artist's cognition, it concentrates the creative process, and that is why it shakes artists, the market, researchers and the public.

Artificial Intelligence (AI) is the field of science that studies, develops and employs machines to perform human activities autonomously. Despite appearing strongly in recent years for having impacted various sectors of society and changing the functioning of countless companies, its studies date back to the 1950s.

The definitive infiltration of Artificial Intelligence in the artistic circuit took place in 2018, the year in which the first AI art was sold for 432 thousand dollars at the renowned English auction house Christie’s. Those responsible for the work were the Parisian collective OBVIOUS, which brings together three researchers, artists and friends interested in exploring the creative potential of artificial intelligence.

Edmond de Belamy (2018) was the work auctioned, and it is the last of eleven portraits that were made for the series La Famille de Belamy, the genealogical tree of a fictional family with classical European aesthetics.

To create this and the other ten portraits, an algorithm was fed with image data from other works of art and existing portraits – created by humans and analogue technologies – and then the machine was trained to create new images from the association of these data. . There are five creation steps: select the subject, data curation, build the algorithm, select the output, that is, how the image should be built, in this case, as a portrait and select the media.
With this case it is already possible to observe some points of discussion. Based on these five steps for creating the image, it can be seen that even if the machine has some autonomy, it still needs human help in most stages of the process.

Artist Bruno Moreschi, in an interview with the newspaper of the Catholic University of Pernambuco (2020), also points out a problem of bias, we often attribute neutrality to technology, however, “at the beginning of AI we had a lot of people with different ideologies and opinions cataloging these images. Indirectly, all labeling, in some instance, was interpreted by humans in precarious working conditions. The field of contemporary image goes through this mediation of a remote worker, who says if this is art or not, if it is important or not, if it is violent or not or even if it is pornographic or not. (…)”.

We also have an authorship problem, after all who is responsible: the collective that trained the algorithm? The algorithm? The programmers? In the case of the portrait, in place of the artist's name was the signature and the equation that was used to generate it.
The problem of authorship also involves the images used in the creation process, an ethical and legal debate. Artists Sarah Andersen, Kelly McKernan and Karla Ortiz filed a class action lawsuit this year in the United States against Stability AI, Midjourney and Deviant Art because their works were used to train robots without their respective authorizations.

https://blog.artsoul.com.br/inteligencia-artificial-arte-artista-meio/