Stories

2022

The project follows different stages of the same process, which are directly illustrated by an interactive interface on the screen. First, the program is responsible for retrieving an image of the person interacting with the artwork. From this photograph, an initial artificial intelligence model is loaded to extract keywords related to the person's clothing style: around fifty results are possible (cotton, graphic, fur, stripes...).

Franck Miquel and I took it upon ourselves to associate each of these words with adjectives that we ourselves have chosen. Next, a second AI model is loaded to create a story based on the following sentence: "Here is the story of a person [keyword]." This story, of variable length, once written, is printed on a receipt ticket using a thermal printer.

Stories generated would then end up facing some absurdity. Most people that got generated their stories found them unusual, strange and dissonant, even though it appeared that some people found parts of themselves in them. In fact, Stories pretends to know you. It is part of these machines that essentialize the human being by pretending that they understand you and that they can give you new information about yourself. By essentializing, I mean that they reduce you to only one of your dimensions.

Choosing clothing as a starting point was a way to base the piece on chosen identity criteria, but actually didn't led it to be more precise. It just remained, as all machines are, biased. We decided, as sensitive humans, that cotton = delicate and tweed = serious. No one knew that the machine worked this way, and so are the ones you are using everyday. Machines are perpetuating, despite them, biases induced by the ones who make them.

In collaboration with Franck Miquel.