The Next Picasso Is a Robot
As the gap between man and machine narrows, it becomes harder to identify what makes humans unique. Robots can perform myriad physical tasks, they can express emotions (even if they don’t actually feel them), and they can learn. So what makes humans special?
Some people think it’s the presence of a soul, though that argument invades sticky philosophical territory and can’t be empirically proven. Cynics might say humans are the only species to make and use weapons specifically to hurt others, but that’s not exactly a reason to boast. Others suggest humans are singular for their ability to make art. From the Mona Lisa to the Taj Mahal, Homo sapiens’ facility for imbuing canvas, stone, sound, and words with beauty and imagination is unparalleled.
Or is it?
Apparently, robots like to rock out. Sound designer Giuseppe Acito put together the LEGO robot Toa Mata Band, which runs on a DIY circuit board-building controller that hooks up to a MIDI sequencer iPad app. The robots use little mallets to strike touch screens and metal keys as the app fills in background beats. The result is a synchronized performance that recreates tunes by bands such as Depeche Mode and Daft Punk.
Alex Allmont’s Play House, an automated sequence of LEGOs, levers, and gears, plays ambient electronic music. This kinetic sculpture combines sequenced beats with sound-producing lever-mechanisms for a multi-layered musical effect. It’s as mesmerizing to watch as it is to hear.
If LEGOs seem too small to jam properly, check out Chico MacMurtrie’s robo-orchestra and art installation Robotic Church. The human-shaped robots are themselves works of art, sculptures made of melted and soldered metal, wires, and pieces of musical instruments. The robots strum the strings that comprise their ribs, hammer away at drums, or dance across a metallic floor. During a performance, MacMurtie conducts the robots, which are powered by cables and hoses. The final result resembles Stomp, but with robots.
Robots aren’t just infiltrating music—they’re writing too. Companies such as Narrative Science have created algorithms turn data into stories. The advantage of these algorithms is speed—they can summarize a football game before the play clock reaches zero. The L.A. Times’ Quakebot can compose an informational article as an earthquake occurs. The Associated Press, Yahoo, Comcast, and Allstate use a platform developed by Automated Insights to turn financial data into news stories at a rate of 2,000 articles per second (and at a cost of roughly $10 for 500 words).
Granted, these stories are all data-driven and lack literary flair, so human journalists still own deep reporting and analysis—for now. Narrative Sciences predicts that work written by its program will earn a Pulitzer Prize any day now, and that computers will control 90 percent of journalism in roughly 15 years. If you’re dubious about robo-journalism, check out this quiz by the New York Times to see if you can distinguish between human and robot writing.
You might be thinking that these examples don’t really count as robots creating art. Humans wrote all of these programs and humans control the robot musicians. Robots can’t demonstrate artistry because the art doesn’t actually come from them, right? Maybe not.
Roboticist Patrick Tresset created a robot that autonomously draws an interpretation of what it “sees” via a camera. Tresset created his first artistic robot, Paul, during a bad case of artist’s block, and has been upgrading the model ever since. Paul IX uses facial recognition technology to accurately represent a subject’s features, but Paul’s paintings aren’t perfect replicas of the models—they reflect interpretation, artistic liberty, and, dare I say, personality. Tresset hopes to demonstrate that robots can autonomously create art that “comments on the human condition.” Perhaps great artists are both made and born?
Google’s Deep Dream Generator uses a neural network (a brain-inspired interconnected artificial intelligence) and its previous experience viewing millions of images to create art. Users can upload a photo and let the generator play with it until it sees something recognizable. The process is similar to us finding shapes that resemble elephants or gnomes in the clouds. This isn’t just art—it’s trippy, dreamy art.
Deep Dream’s imagery has become so popular that a San Francisco art gallery recently hosted a show featuring works from ten artists and engineers whose creations demonstrate the art of neural networks. At auction the pieces brought in nearly $100,000.
While it might seem tempting to pit man against machine to determine artistic mastery, perhaps the better approach involves combining human and AI skills, allowing real and artificial neural networks to flex their creative muscles in tandem. As they say, if you can’t beat ‘em, join ‘em.