Google is planning to put the art in artificial intelligence. The announcement was made at a session in Moogfest, a four-day music and technology festival in Durham, North Carolina.
The company’s artificial researcher Douglas Eck had developed a new group that would focus on figuring out what computers can truly create. The group is named Magenta, and they will launch more publicly at the start of June. However, the Moogfest attendees got an introduction on what it will be working on.
According to the company, Magenta will be using TensorFlow, which is the machine-learning engine that Google built and opened to the public at the end of 2015. It was launched with an aim to know whether AI systems can be trained or not. It was tested to create original pieces of music, art or video.
However, it is not an easy task as even the most advanced artificial systems are unable to copy the styles of existing artists and musicians. Eck conceded that the AI systems are “very far from long narrative arcs” during the panel discussion.
However, Magenta will help create tools for other researchers as well as for its own team and get to know the creative potential of computers.
Similar to the launching of TensorFlow, Eck said Magenta will follow suit. First of all, the company would be launching a simple program that will help researchers import music data from MIDI music files into TensorFlow. This, in turn, would help their systems get trained on musical knowledge.
A member of Eck’s team, Adam Roberts told Quartz that the June 1 Magenta group will be posting more information regarding the resources it will be producing and adding new software on its GitHub page. Along with it, the team would post regular updates on a blog.