Machine dreaming – GANscapes is a series of five animated digital collages based on images generated with artificial intelligence (AI) that work with generative adversarial networks aka GANs
These networks allow the AI to generate realistic images (or sounds, voices, music) starting from a given “dataset” and via machine learning. The interesting thing is that you can ask a GAN to generate images that are a mix of categories, pushing the boundaries of the human imagination.
The result are anomalous images that look like something but are not like anything, made up of absurd and alienating objects that do not fit into any human taxonomy. Rather, they belong to a kind of machine unconscious capable of generating dreamy compositions put together by the GAN algorithm, whose parameters have been altered.
I intervened on these scenarios, which I call “GANscapes”, with some details, dosing my collaboration with AI. I therefore created five “rooms” to see, from inside, how a machine dreams of us. Each room hosts the voice of digitally GAN generated famous singers: Mina, Frank Sinatra, Ella Fitzgerald, David Bowie, Elvis Presley.