Dancer, roboticist and mechanical engineer Catie Cuan and a UR5e robot arm. (Photo by Adam Lau/Berkeley Engineering)Bridging the brain
In the performance piece “Breathless,” a UR5e robot arm flails wildly back and forth, as dancer and roboticist Catie Cuan arches her full body to mimic its movements. The unnatural effect of the robot’s motion becomes that much more heightened when paired with the human body. But Cuan never offers a 1:1 copy; the eight-hour piece aims to spark conversation about the relationship between artificial intelligence and the future of human labor.
So it’s fitting that she gave a presentation on this collaboration with Ken Goldberg, professor of industrial engineering and operations research and of electrical engineering and computer sciences, in the latter’s class, Beyond the Uncanny Valley: Art, AI and Robotics.
A science fiction staple, the “uncanny valley” theory posits that as robots acquire more human-like features, they become more likable — until they don’t. That “valley” was charted by Japanese roboticist Masahiro Mori in 1970 to describe the frightening sensation of observing objects that are too human-like.
The course, co-led by Lisa Wymore — professor of theater, dance and performance studies — seeks to offer historical context to bridge the artistic and technical understanding of technological advancements in the wake of the AI explosion. The syllabus covers everything from dance to deepfakes, with plenty of insights from guests working at OpenAI, Niantic and the Department of Art.
“I was really interested in teaching students about the history of ‘uncanny’ that predates the ‘uncanny valley,’” Goldberg said. “I would say most engineers don’t know that background.”
On the flip side, Goldberg also intends to provide context to humanities majors who “don’t know the history of AI.”
This left brain-right brain approach is seeded into the class structure. One minute, students can hear graduate student instructor Curtis Rumrill of the music department unpack his unsettling composition on the violent lifespan of a farmed rabbit. The next, graduate student instructor Simeon Adebola of the AUTOLab might explain in detail how machine learning helps robots move.
Goldberg sees this interdisciplinary format as a reflection of the zeitgeist. “These fields are starting to come together in interesting ways,” he said. “That’s really exciting, because it brings together viewpoints. It’s intellectual diversity. Engineers think a little differently than English majors.”