Michelangelo’s “Creation of Adam,” processed through Google DeepDream (image via Kyle McDonald/Flickr)

Can artificial intelligence create art? In recent years, we’ve seen plenty of attempts to answer this question, from portrait-painting robots to Aaron, a computer program that makes neon abstractions. Now, Google is launching Magenta, an effort to generate original music, video, images, and text using machine intelligence. Part of Google Brain, the company’s deep learning research project, Magenta will explore whether artificial intelligence can create art.

“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. And I wanted to demystify this a little bit,” Google researcher Douglas Eck said during a panel at MoogFest, a technology and music festival. He was inspired in part by Deep Dream, Google’s computer vision program that turns digital images into psychedelic nightmares.

“The question Magenta asks is, ‘Can machines make music and art? If so, how? If not, why not?’ The goal of Magenta is to produce open-source tools and models that help creative people be even more creative,” Eck wrote in a blog post. “I’m primarily looking at how to use so-called ‘generative’ machine learning models to create engaging media.” Today, the researchers published the first of their results on TensorFlow, Google’s open-source artificial intelligence platform. The first stage of the program helps researchers import music data from MIDI music files into TensorFlow, which will, they hope, imbue their systems with new musical knowledge. After experimenting with music, researchers will move on to images and video.

YouTube video

The definition of “art,” of course, is always up for debate, for better or worse. Whatever Magenta ultimately creates, it will no doubt reignite this age-old conversation. Eck says his goal with Magenta is to give listeners “musical chills” with an AI-composed piece (since, you know, human musicians aren’t struggling enough as is). But even if Magenta succeeds in producing a robot Beethoven, can an algorithm be considered an “artist” if it was designed and enacted by a human? Until computers are creatively independent and display artistic intent, they’ll likely be considered just another new tool for human art-making practices, not artists in their own right. Despite Google Deep Dream’s masterpieces, computational creativity hasn’t come close to producing its Sistine Chapel. Eck admitted at MoogFest that algorithms are still “very far from [understanding] long narrative arcs.” Maybe future generations of artists will be competing with bionic men for gallery shows, but don’t panic about the robot art world takeover just yet.

h/t QZ

Carey Dunne is a Brooklyn-based writer covering arts and culture. Her work has appeared in The Guardian, The Baffler, The Village Voice, and elsewhere.

2 replies on “New Google Project Aims to Put the AI in “Artist””

  1. Meh….I tried it…it just keeps putting canned stuff in the file and distorting it – lots of animal eyes. Filters on filters on filters…blargh….
    To me Real AI would be that I describe to the computer what I want the art to look like; such as “Draw a cat in a tree next to a pile of cheese that looks like Marilyn Monroe…all in the style of Picasso and Van Gogh” and then the computer would make it and adjust it as I tell it to make changes.
    (And then, computer, if you are so smart, sell the art for me.)

Comments are closed.