Google's latest project, Magenta, plans to train AI systems to make art, music and videos
We’ve seen AIs writing poetry, finishing the works of dead authors and philosophizing over the mean of life. Now the plan is to put the art back into artificial intelligence.
As of June 1, Google will launch Magenta, a project which will work on teaching artificial intelligence to create art. The research team will use TenserFlow, a Google-built, machine-learning engine that was shown to the public last year. They’ll also be working on its ability to create original pieces of music and video, as well as art.
Speaking at Moogfest, a music and technology festival in North Carolina, Google Brain researcher Douglas Eck outlined that their biggest problem was the AI system’s current inability to not copy existing artists. He said the current systems are “very far from long narrative arcs.”
The project hopes however to rectify this, opening the Magenta project to the public to allow the system to further expand its knowledge. Eck explained that Magenta was inspired by other Google Brain projects like DeepDream, where the AI learned to fill in the gaps with images, creating trippy, psychedelic structures as it guessed what would naturally come next.
“There's a couple of things that got me wanting to form Magenta, and one of them was seeing the completely, frankly, astonishing improvements in the state of the art of creative deep learning. And I wanted to demystify this a little bit,” Eck explained.
There was also talk at Moogfest about launching a Magenta app alongside the project, to gauge whether the general public liked the art the AI was producing, or whether they were just caught up in the novelty.