Browse through photos of never-before-seen art at This Artwork Does Not Exist. The site was created by Philip Wang, and uses a generative adversarial network (GAN) to create images of artwork that don't actually exist.
Generative Adversarial Networks, which were invented by Ian Goodfellow and his colleagues in 2014, use machine learning to produce content. GANs have the potential to create a realistic-looking version of almost anything. To see a new image of non-existent artwork, just refresh the page. The results are never-ending.
All of the images I came across on the site are non-representational and pretty similar in their use of shapes and their color palette. I find the images to be pretty monotonous after a while of refreshing the page (they could all be mediocre café wall-art), but the technology excites me because I imagine that as the AIs get more sophisticated, we'll see more variety and more representational artworks as well.
This technology's ability to instantaneously generate an image that could easily be a human-made abstract painting, makes me ponder the age-old question "what is art?"