Electronic Arts

treads the fine line between man-made and machine art

dream 967d7cf6b4(1) In his aptly named essay What is Art?, Leo Tolstoy proposes a definition of art as “a means of union among all men, joining them together in the same feelings, and indispensable for the life and progress toward well-being of individuals and humanity”. A grand definition, to be sure, and one which attempts to pin down a famously nebulous concept. Art exists in many forms, but it was the actions of a robot made by one of Google’s Creative Lab teams that provoked vitriol from Jonathan Jones in The Guardian after this year’s Mobile World Congress.

The robot is actually just an Android phone (the Nexus 6P, for those playing at home) connected to a microcontroller, which in turn controls a pen suspended by wires to ‘sketch’ a portrait from a photo taken by the phone. “Lo and behold,” Jones cries, “the art of portraiture has now been replaced by an app and a robotic arm.”

Jones’ statement is, of course, wrong, and he argues as such. In fact, I could find no evidence that there were any claims by Google or the team that their robot was indeed creating art, or doing anything more than using darkness and contrast to trace a line between points on an image which just so happens to be of your face. They could just as easily have taken a photo of a page of text, but that probably wouldn’t have been quite as exciting as photocopiers have existed since 1938.

But the very fact that Jones saw fit to argue against the idea that this robot is creating art raises several interesting questions about the nature of art itself, namely whether one needs to be human in order to create art.

Nowadays the answer to this question is far from clear-cut. On one level, this robot is little more than a glorified paintbrush: a tool used and created, fundamentally, by humans in order to produce an effect that would be laborious or impossible without it. This robot does just that, by taking an image (captured using another man-made tool) that has been human-composed, and translating it into a line by looking at contrast in that image. In this way, the robot is indeed just a “high-tech Spirograph toy,” as Jones puts it, and renders any sort of comparison to Rembrandt frankly ridiculous. After all, computers can only output based on their input, and for such a simple (relatively speaking) process as this, that input is unavoidably human.

Decidedly un-artistic photos can be made into stunning psychedelic masterpieces, at once specifically computer-created and utterly unpredictable

But what about more complex processes? My mind jumps to another project of Google’s, namely ‘Deep Dream’. Based on the company’s machine learning AI software, it takes a photo and analyses it to find similarities with other photos it knows, in the same way that Google’s ‘similar image’ search works. This works well when it is told to find a dog in an image of a dog, for example, but if the image has no dogs the program instead finds the most dog-like thing in a picture of, for example, the Manhattan skyline. Once it does, it changes that aspect to look more like the thing (in this case a dog) it looked like, and runs the image through again. Applied to every image the program knows (not just the pictures of dogs, although animals are common), and combined with a kind of feedback loop, strange and surreal images can be made, automatically, out of seemingly mundane pictures. Suddenly, decidedly un-artistic photos can be made into stunning psychedelic masterpieces, at once specifically computer-created and utterly unpredictable.

Now, arguably the programmers are the artists here. After all, they made the program, and ‘taught’ it to recognise qualities of different subjects by feeding it images. But the complexity of this process blurs the line significantly. We may afford great respect to the tutors and masters of great artists, but we still consider the artist’s work to be their own. Add to this the fact that the images Deep Dream creates are not only impossible for a human to accurately recreate in the same way, but are only possible for a computer to create, and the argument that art has to come from a human mind gets rather more shaky.
This is not to say that human artists will all have died out in the future, or that before long computers will be accurately replicating Van Gogh. But the potential complexity and processes to which computers are privy means that we are very close to being unable to treat them as just another tool in the artist’s studio.

One comment

  1. If you think googles sketch robot is interesting, check out robotart.org where there is half million dollar competition for robots to paint paintings. If you have one, there is still time to register.

    Reply Report

Leave a comment