It’s been said that art is in the eye of the beholder. Now, computers are able to predict when artwork is appealing to the eye and the brain.
Whether it’s the abstract images of Jackson Pollock or the natural light of impressionist painter Claude Monet, computers now have the power and ability to determine art’s subjective appeal. Scientists at the California Institute of Technology pulled this off with some old-school technology, using surveys to get 1,500 people to rate four styles of art — impressionism, abstract, cubism and color field.
Then, they deployed a computer to break down a painting into its low-level visual elements such as color saturation, hue and contrast. They also tasked the computer with assessing the paintings’ “high-level” features, such as whether a painting is still or dynamic. These high-level features are where human-driven judgments about aesthetics come into play.
After some “learning” time, the computer program was able to accurately predict which paintings people would like. The scientists said the findings shed new light on how people make aesthetic judgments. It also quantifies a way to predict what had been considered a highly subjective decision — an artwork’s appeal or lack of appeal.
The program works by estimating how much of the various individual features a person considers when evaluating a painting. The researchers also discovered that many people prefer colorful abstracts, the complexity of cubism or “real-life” paintings such as impressionist works.
So whether your taste runs toward Pablo Picasso or Georgia O’Keeffe, there’s a computer out there that knows.