How would you engage with an image if you knew that it was the last memory of a place that has filled your heart with peace for many years? Refik Anadol’s Machine Hallucinations: Last Memory is a data sculpture that explores this feeling by making a machine mind “hallucinate” vestiges of nonexistent natural wonders. After the machine learns 46,474,696 million images of Nature, Anadol uses GAN algorithms developed by artificial intelligence to shape this dataset through his unique artistic vision. In addition to utilizing cutting-edge technologies to create an artwork that commemorates the beauty of this land we share, the artist speculates a future inflicted with the danger of being left only with the memories of Nature. The immersive piece incorporates earthly pigments, shapes, and patterns that we associate with our sensory experiences while paying homage to the Earth’s unbound poetic sublimity.

Sound Design : Kerim Karaoglu

This AI artwork trained with StyleGAN2ADA algorithm developed by NVIDIA and latent walk generated by custom software developed by RAS.

For more information :
Training Generative Adversarial Networks with Limited Data
Tero Karras, Miika Aittala, Janne Hellsten, Samuli Laine, Jaakko Lehtinen, Timo Aila
arxiv.org/abs/2006.06676