In October 2018, a portrait produced by artificial intelligence (“Edmond de Belamy, from La Famille de Belamy,”) sold at an auction at a price before fees of $350,000.
Some consider this a pivotal moment in the future of AI and art. AI has exploded over the last few years with faster and cheaper computers allowing for deep learning to develop in more advanced ways.
The concept of AI has been discussed for generations — even before the robots in Star Wars and the Jetsons there was AI science fiction in the 1927 movie Metropolis.
AI art is produced by Generative Adversarial Networks (GAN). Without getting too technical (here is a deeper explanation if you wish!), the AI is fed loads of reference artwork and the AI figures out how to emulate it. AI can start to categorize certain art styles and imitate those same styles in slightly different ways.
This year, Nvidia demonstrated how it can turn simple sketches into images that are nearly photorealistic with a software called GauGAN. Another program “AICAN” can create art that 75% of people think was created by a human.
Artists shouldn’t be afraid of their jobs, however. AI is still more about imitating human artists than being truly innovative with new creative styles. AI currently has issues with intuitively understanding social value in art. The human artists are very present in the AI’s artwork. Similar fears plagued artists in the past with the mass adoption of photography.
Besides the scientific and social challenges, there are legal challenges to be considered as well. Ownership issues with the commercial nature of AI created products. Billing issues of AI developed art. Would an artist’s copyright go to the AI if the AI is loosely referencing a hundred different artists?