Christie’s became the first auction house to put an AI-generated artwork under the hammer. Created using a generative adversarial network (GAN) trained to make new images based on a dataset of 15,000 portraits painted between the 14th and the 20th centuries, the piece sold for $432,500 - over 40 times its $10,000 estimate - prompting widespread speculation about whether AI could be behind the next great art movement.
Although there’s a 30-year history of artists employing algorithms and computation within their practices, this spike in contemporary AI art, led by artist/software engineers has occurred in the last five years or so, made possible by recent breakthroughs in machine learning like GANs, which, by pitting two neural networks (machine networks that approximate how the brain works) against each other, can be used to create original pieces that are increasingly difficult to tell apart from those made by humans. Also vital to this development is Google’s DeepDream, invented in 2014, which uses a convolutional neural network to find and enhance patterns in images.
Is AI art any good?
But while technically impressive, are these artworks actually any good? Well, it depends on who you talk to. At this year’s AI-themed Christie’s Art + Tech Summit, Jason Bailey, founder of art analytics website Artnome, argued that if what separates our generation from its predecessors is the invention of computers, then software or computational art is necessarily the most important art of our generation. In theory, his statement makes a lot of sense, but faced with paintings of blurry-faced, GAN-enabled portraits or the psychedelic landscapes produced with DeepDream, I can’t help thinking that we’re not there yet.
It is a creative tool that can easily help artists work and provide artistic inspiration. It creates a work of art of a specific art trends, converts it into the style of a certain artist, and provides a service to change the style of the art work.