Take Five: Artificial Intelligence and Art History

The Washington Post recently printed an article describing an algorithm that Ahmed Elgammal from Rutgers University used to identify the creative worth of a work of art. Working at the school’s Artificial Intelligence Laboratory, Associate Professor Elgammal devised a system of mathematical transformations to identify creativity.

Elgammal explained, “Through a series of mathematical transformations, we showed that the problem quantifying creativity could be reduced to a variant of network centrality problems – a class of algorithms… Any algorithm’s output depends on its input and parameter settings. In our case, the input was what the algorithm saw in the paintings: color, texture, use of perspective and subject matter. Our parameter setting was the definition of creativity: originality and lasting influence.”

The experiment, that analyzed 1,700 painting throughout history, yielded some impressive results, positioning works of art at the top, such as Pablo Picasso’s Les Demoiselles d’Avignon and Edvard Munch’s The Scream, that most art historians would agree are profoundly original and influential works of art. The goal of such an experiment, which was to create a machine that has the perceptual and cognitive abilities of a human, seems to have been achieved if one is evaluating the “creativity” of white European and American men working during and after the late 19^th century.

The experiment failed to properly evaluate all other works of art because of the parameters that Elgammal set. No western civilization highly valued originality in art before the late 19^th century. In fact, to the contrary, during the Renaissance, art was appreciated for its references to Ancient Greek and Roman art. During the 18^th century, art that quoted Ancient and Renaissance paintings, sculptures, and prints was highly esteemed. Furthermore, the vast majority of art produced before the 20^th century was commissioned and therefore the patron determined the subject matter. It is perfectly correct that Leonardo da Vinci’s Mona Lisa should not rank highly in this experiment given the parameters, but it is incorrect to conclude that the work of art does not have a high creative worth. The standards were different in earlier centuries.

Finally, the lasting influence of a work of art largely depends upon its visibility. Simply put, this was determined at least in part by an artist’s race and gender; therefore, there are many quite creative works of art by women and minorities that wouldn’t have the same influence because they are less visible.

Artificial Intelligence could be more thoughtful.