With an increasing trend in the use of Generative Adversarial Networks (GANs) comes an intriguing new age for the art world.
It has become a popular topic amongst the mainstream media that whilst the use of artificial intelligence is gaining traction, should humans fear being challenged or replaced in the workplace? Maybe that fear is soon to be tested.
In October of this year, Christie’s auction house in New York sold a piece of artwork for $432,500 labelled ‘Portrait of Edmond Belamy’. He’s a contemporary, aristocratic-looking gentleman surrounded by a unique combination of both dark brushstrokes and negative space. His indistinguishable facial features at first glance seem reminiscent of a hazy dream you would have seeing many faces in a busy crowd. Believe it or not, those brushstrokes were imitated by artificial intelligence technology.
French art collective Obvious have been testing the boundaries of what art really is with exhibits featuring other creations of theirs that have a likeness to Edmond Belamy. The algorithm Obvious uses are split into two parts, one being the Generator and the other the Discriminator. The machine’s dataset carried more than 15,000 portraits, varying from the 14th century to the 20th. The Generators job is to create a new image based on the data provided. On the other hand, the Discriminator tries to separate the differences between human-made artwork and the Generator’s depiction. If the Discriminator is fooled into thinking the Generator has made a real-life image, then the algorithm is successful. Following this ‘Gan-ism’ method has helped Obvious make another 10 portraits that sit proudly alongside Edmond in a family tree styled theme.
In New Delhi this year, the Nature Morte museum hosted a vibrant selection of works named Gradient Descent from seven different artists who have harnessed machine learning to become some of the worlds AI-Art firsts. It is fascinating to see these exhibits side by side, as each of the installations vary so greatly in both technique and style. Using Gan-isms to form art is a promising way to look at artificial intelligence. Unlike the initial revelations that came with Google’s DeepDream which created a twist on pre-existing images, Ian Goodfellow’s essay from 2014 stated that the concept of Gan’s could formulate its own entirely new product.
Of course, not everyone agrees that manipulating artificial intelligence for creative purposes could ever truly make art. It lacks the human emotion and experiences that a real-life artist would have endured to inspire such works… Something that a machine could never really embody. There are also other artists in this particular field who don’t agree that Obvious are accurately representing them as the Belamy collection is merely a replication of art that already exists. It’s not that original. However, it is worth noting it bears resemblance to other portrait art in the field such as Mario Klingemann due to the general lack of structure and distortion in the facial features. This is perhaps because this technology is yet to learn the curvatures and complexities of the human face, so those areas remain a little fuzzy.
Whilst we must remember AI-Art is still very much in its infancy, it is exciting that the discussion of something that for so long has been archaically known as intangible and rigid, may now begin to take strides into unknown realms of possibility.