Sergio Monje

From hundreds of postcard photographs of the Balearic coastline from the Planas Archive, we have trained a Generative Adversarial Network (GAN), an artificial intelligence tool that allows us to generate new images of imaginary littorals, new (im)possible coastlines. Thus, we poetically trace alternative streams of the history of the Balearic landscape, producing a new parallel archive to that of Casa Planas, formed by hundreds of artificial and unreal postcards of a non-existent archipelago.

If we live immersed in what we could define as a touristic realism (similar to Mark Fisher’s capitalist realism, that inability to imagine alternative models to capitalism that permeates the culture of recent decades), can assisted imagination help us escape from it?

It is possible, but a machine learning system always starts from a fundamental limitation: its capacity to invent new images is subordinated to the material that forms the dataset with which it has been trained. What if it turns out that our imagination, like that of a neural network, is also limited by the torrent of images that saturates the imaginary? Is not the imagination of touristic realism analogous to that of machine learning? How can we get out of the closed and repetitive loop of infinite variations of this machine imagination, to open space for difference? That is: how can we exceed touristic realism?

Thanks to:

Alelí Mirelman, Marina Planas, Dora García, Begoña Márquez, Juanmi Bauzà, Anna Berenguer, Marijo Ribas, Copycorner, Xicarandana

With support from: