AI Can Now Produce Better Art Than Humans. Here's How.

An AI Picasso

Scientists are using artificial intelligence (AI) to find a "new system for generating art" and testing their results on the public. The system, called a generative adversarial network (GAN), works by pairing two AI neural networks: a generator, which produces images, and a discriminator, which judges the paintings. It does this based on the 81,500 example paintings and knowledge of different artistic styles (such as Baroque, Impressionism, and Modernism) it was taught. The suggester creates an image, the discriminator criticizes it, and the conversation leads to a work of art.

The scientists changed the way that AI usually produces art by having the generator only create works that did not fall into a preexistent category of painting — they did this by "maximizing deviation from established styles and minimizing deviation from art distribution," according to the abstract.

 

Mark Riedl, an associate professor at the Georgia Institute of Technology in Atlanta, said that he liked the "idea that people are starting to push GANs out of their comfort zone — this is the first paper I've seen that does that."

After the paintings were produced, the scientists conducted a survey with members of the public in which they mixed the AI works with paintings produced by human artists. They found that the public preferred the works by AI, and thought they were more novel, complex, and inspiring.

The Work of Art in the Age of Mechanical Creation

Paul Valéry, who Walter Benjamin used as a starting point for his essay "The Work of Art in the Age of Mechanical Reproduction," wrote in 1931: "We must expect great innovations to transform the entire technique of the arts, thereby affecting artistic invention itself and perhaps even bringing about an amazing change in our very notion of art."

He was referring to the modernist period, in which new techniques and ideologies changed the way art was perceived. We may be experiencing a similar upheaval in the art world. Benjamin's criticism of the exact copies that could be produced by the second half of the 20th century centered around the idea that "even the most perfect reproduction of a work of art is lacking in one element: its presence in time and space, its unique existence at the place where it happens to be."

This AI project possesses this property. It does not just copy or manipulate, as Google Deep Dream does, but is able to produce true works of art by being actively programmed to be novel and creating originals in a specific place. These pieces are more similar to Aiva, an AI composer that also could not be detected by humans, than it is to Deep Dream.

We are entering an age where AI is becoming increasingly ubiquitous and competent in almost every field — Elon Musk thinks it will exceed humans at everything in by 2030 — but art has been viewed as a pantheon of humanity, something quintessentially human that an AI could never replicate.

Studies such as this show that our artistic leanings may not be off limits — and with AI conquering humans at our own games, like chess — how long is it before we create a Picasso program that is superior to any current human artist — and immortal to boot?