Google's Project Magenta wants AI to create original art as well as tech robots to make music

Google demonstrated Project Magenta AI that can generate music and other art at a technocultural event, Moogfest.

Google previewed attendees of the Moogfest, a technocultural music and computation festival with a presentation on Project Magenta. Project Magenta is going to use Google's TensorFlow, an open source distributed computing machine learning and neural network framework. Project Magenta is going to use generative algorithms to make music as a priority. After that it is going to create videos and images as well.

A part of the demonstration was a researcher playing notes, and the machine picked up those notes and used them to form a more complete melody. Project Magenta will get a public launch in early June. A smartphone application launch is also in the works. This application will allow users to browse through artistic creations of Project Magenta.

Machines have been used to make art, especially procedurally generated art since the early sixties. The Spirograph, where geared circles with holes in them are used to generate patterns can be considered a very basic form of computational art. But these involved some kind of coding from a human, with some idea or expectation of the final product.

Researchers  from Microsoft and Dutch cultural institutions have created a Rembrandt 350 years after his death, using machine learning and a 3D printer that layers on the paint carefully to imitate the brush strokes of the master. Of course, this artifically created painting wasn't liked by the critics and challenge was raised as to whether it can be considered to be a piece of art. 

But, the kind of art imagined by Google's Project Magenta is to make the process of creation totally independent of human agents, so that the machine is doing all the work. Google has previously shown what the dreams of learning machines look like, where they made machines create recognisable pictures over repeated iterations.

More details from the announcement at Moogfest are at Quartz and Tech Times. Those interested in diving in deep can visit the GitHub page for Project Magenta.

Find our entire collection of stories, in-depth analysis, live updates, videos & more on Chandrayaan 2 Moon Mission on our dedicated #Chandrayaan2TheMoon domain.