ciel = terre is an online AI artwork dealing with human myths on machine learning for creativity.
In terre, the spectator is virtually put inside a cube, whose six faces successively display experimental music videos. Interactivity, often erected as an innovative feature of AI, is brought back to the material experience of traditional audiovisual media: staring at a movie screen, or listening to music in one’s own soundscape.
In ciel, the spectator is presented with an oval map, whose surface continually streams neural noise from an audio VAE and an image GAN. The modern myth of unlimited mass production, fuelled by mainstream streaming platforms, is sublimed by the stardardization, anonymization, and automation of media creation enabled by deep neural networks.
Intimate audiovisual data, archived with my Blackberry camera between 2011 and 2018, served as material for composition in terre, as well as for training and generation in ciel. The power of machine learning technology to harm or sustain human creative cultures is = to the personal myths, dreams and beliefs of the living communities who appropriate it.