A tech professor is leading a research team focused on teaching artificial intelligence to dance, and they plan to put on a dance performance with their creation at the end of the project.
Brian Magerko, professor of digital media, works at the heart of computation and creativity, and his new project seeks to unite AI and entertainment.
Magerko explained how the latter company builds on its previous ventures.
“This is an extension of a project that has been going on for eight years now,” Magerko said of the depth of his research.
“The idea was to build an AI that could learn to dance by dancing with people,” Magerko said, and this project resulted in the creation of LuminAI.
This AI was able to perceive people’s dancing and gestures and imitate their dance moves, and even improvise based on certain moves she had already learned.
Magerko and others have shown this technology around the world, hosting many interactive exhibits with AI.
His experiences with LuminAI inspired Magerko to continue his research on this new project.
“Thanks to our iterative work, we are developing[ed] some theories about how, at least in an embodied sense, people improvise and how computers could improvise with us in a more fluid and formal way, ”Magerko said.
He pointed out that understanding how humans dance and create movement is key to applying something similar to AI.
“This will inform us about how bodies make sense in relation to each other,” Magerko said, “and that’s an important thing for robots and intelligent agents.”
This knowledge would have much broader applications than the world of dance and would further help cement artificial intelligence as the next front in the quest for scientific exploration and discovery. For all of his technological applications, however, Magerko also admired the opportunity to put some form of beauty into computing, transcending the line between code and art.
Similar to the LuminAI project, the results of this research will be presented in a public performance.
Magerko said the team plan to showcase their AI in a collaborative performance with the dance department at Kennesaw State University upon completion of the project. To many, this may seem like an area AI would only reach in a dystopian sci-fi movie.
Still, Magerko has managed to push the boundaries of AI enough to bring this technology into the present, with a live performance slated to showcase AI’s capabilities and talents.
Magerko said the performance will likely occur “using motion capture suits as a mechanism for AI to sense, react and dance with performers on stage in real time.”
It’s not sure how this AI will be represented during the performance, but they can take advantage of the technology they used to present the LuminAI.
The project is expected to last three years, the first year being devoted to studying the movement of dancers.
Part of this will include feeding AI examples of dance moves and what humans think go well together, and this method will be accentuated using dance theory to teach it. In doing so, Magerko hopes to create an AI capable of recreating modern dances at a level close to, if not perfectly, indistinguishable from those created from a human mind.
“One of the things that helps machine learning work more efficiently is good descriptors for your knowledge or data,” Magerko said.
Labeling consistent movements in the dance would make the machine learning process much easier, according to Magerko.
He also said that the AI would mainly learn contemporary movements based on more modern dance forms for the time being.
Magerko said that the second year of the project will be devoted to building models based on the data collected, and the third year will be devoted to maturing and evaluating the work done and implementing actual performance.
He is eager to put on the show and prove that “we can improvise with computers with our bodies in a very naturalistic and coherent way, improvised”.
The show, it seems, is set to become another milestone not only to prove what AI is capable of, but also to question its limits, if any, to learn and emulate what was once considered a behavior that only humans could achieve.