“When I’m searching for new sounds and new robots, I’m thinking of like, I need this sound in music,” Geist said. “If I would make my track in the computer, I would need like a hi-hat or whatever. So I’m thinking, how could this be done in the real world? OK, a noise like tshh. How does this happen here? And then you easily come to pneumatic air-pressure things.”

“Have an open eye to the world and see structures, get inspired, connect the dots.”

The robots have secrets of their own. It takes Geist an average of four months to put together a robot for touring, which includes instruments made out of printed circuit boards, 3D-printed kalimbas, flashing LEDs and pneumatic valve systems. Each robot is custom, and they all have visual elements of some kind. The air-pressure machine that serves as a hi-hat, for instance, includes a series of tubes filled with small white styrofoam balls that bounce up and down with the air flow. At first blush, it looks like the tiny spheres themselves are making the tshh noises, but listen closer, and it’s clearly controlled bursts of air pressure.

“I think it’s very important to just have an open eye to the world and see structures, get inspired, connect the dots,” Geist said. “See structures where you think, OK, I could really use this and just like turn this into a metaphor for whatever. This is super important.”

Geist, who teaches a class on the evolution of technology and society at NYU Berlin, is convinced robots will only become more integrated into the music industry as time passes, just as they’re infiltrating wider society. Robots and artificial intelligence are tools, he said, and like any tool, they’re neither inherently good nor bad. It all depends on how we use them.

“As humanity, we started off with hammers and really crazy stuff to extend our capabilities of a human being,” Geist said. “I mean, robots and artificial intelligence — the robots are made for the body and the algorithms are made for the brain, so there’s this duality. This is an extension of our capabilities in the world and the interaction between us and the world, physically and metaphorically.” He paused and laughed. “What was the question?”

“Humans are very, very good in detecting the smallest, subtle changes.”

Geist’s coming EP is fittingly called Speculative Machine and it’s a soothing, booming, four-track exercise in resonant bass and tinkling metal. Even as he helps craft the AI-powered takeover of the music industry, Geist isn’t concerned about the robots he builds taking his job. Already there are programs that will generate copyright-free songs for people to use in personal projects, podcast intros, Kickstarter campaigns or trailers — functional music. However, music is as much emotion as it is carefully placed rhythm, and a song with any staying power will require that human connection, Geist said.

“The step from functional music to music which is emotionally connected to this, kind of like pop music, that’s a big leap,” he said. “It might take a while until this is replaced. I don’t think it’s so easy, because in the end, humans are needing something they can connect to emotionally. For me, AI is very difficult to do this in a whole in the pop music scene. … This emotionality — humans are very, very good in detecting the smallest, subtle changes.”