The RoboThespian, a human-size robot developed by Engineered Arts, based in Penryn, England, often works as a museum guide. It has also done some very stiff stand-up comedy and a slightly less stiff rendition of “Singin’ in the Rain.”
But these are largely solo performances. In July, it did something more unusual: It performed alongside two humans in a play called The Uncanny Valley, which made its New York City premiere at the Brick Theater in Brooklyn. (Clearly, robots are in the zeitgeist, as another production called Uncanny Valley played in New York around the same time; its robot character was portrayed by a human actor.) Both plays are named after the distressing psychological gray zone that people experience when faced by a robot or avatar that looks and acts enough like a human to be creepy, but not enough to be convincing.
The Uncanny Valley is set in a university laboratory. The RoboThespian plays Dummy, a research robot connected to a vast computer bank. Dummy works with a human research subject named Edwin. Every time Edwin visits the lab, he sniffs a vial and dons a sensor-laden skullcap. Each odor triggers memories and emotions, which Dummy absorbs, and gradually, the robot’s voice and face start to resemble Edwin’s. Dummy tells Edwin that the researchers in charge are working on a lifelike avatar, with biosynthetic skin and realistic facial expressions. Soon, Edwin is telling Dummy that someone who looks just like Edwin is walking around town, going to his job, and talking with his friends. In no time at all, it’s looking like Edwin will be replaced.
Although the robot stayed seated throughout the play and was rather limited in its body language, it didn’t seem stiff—most likely because the performance relied heavily on human beings. With the help of a US $50,000 grant from the University of North Carolina at Chapel Hill, creator and director Francesca Talenti and telepresence researchers at the university pieced together RoboThespian’s performance from a combination of motion-capture data, voice recordings, and video of actors’ faces. An internal rear projector was used to cast this video onto the interior of the RoboThespian’s translucent white mask.
The motions, sounds, and video were divvied up into segments that were triggered during the play by a “puppeteer” on an offstage laptop. In some longer segments, the robot would perform without intervention, and it was up to the actor playing Edwin to make sure he timed his lines properly. “The actor had to be very consistent in order to not trip up the robot,” Talenti says.
Despite the RoboThespian’s voice and face becoming more like Edwin’s, the play never really gets close to the depths of the uncanny valley. RoboThespian never really leaves the obviously mechanical side: There’s only so much you can do with a spindly-looking roboskeleton that hisses seemingly every time its pneumatic actuators engage.
That said, Dummy does start to seem more human over the course of the play: It experiments with jokes; it gushes about its pet cat. At times, the robot’s distorted, video-projected face makes it seem as if there’s someone inside the robot, trying to escape. I got that feeling especially toward the end, when Dummy realizes the research project has done Edwin harm. The thought causes the robot pain, and it cries out in anguish in Edwin’s voice.
This isn’t what usually happens in science fiction when a robot becomes as smart as we are. And it raises a troubling possibility as the play nears a resolution. Has a new, sentient being been brought into the world? Or is the machine just playing out an empty model of human behavior? It’s the Turing test with life-and-death stakes—and a conundrum we hopefully won’t have to tackle for a long time to come.