Google’s Project Magenta Turn Machines Into Artists, Musicians

Updated on

Git Hub, the “social network” turned code repository, has received a gift from the Google Brain team of tools and open source data models in order to see if through machine learning computers might be able to be come painters, musicians and artists in other mediums.

Google’s TensorFlow at work with Project Magenta

The Google Brain team named its machine learning framework TensorFlow in November of last year and this is just one example of the company asking others to help them with developments in machine learning and by proxy artificial intelligence. In the case of Project Magenta, the company is curious if a machine can create art that is deemed just that by humans. Surely, we’ve heard how Google and others have been working on incorporating machine learning to understand things in context rather than simple speech recognition, but to ask for an original piece of art is another animal altogether.

“Machine learning has already been used extensively to understand content, as in speech recognition or translation,” said Google research scientists Douglas Eck in a June 1 blog post. “With Magenta, we want to explore the other side — developing algorithms that can learn how to generate art and music, potentially creating compelling and artistic content on their own.”

Perhaps, you’ve heard the expression that (I paraphrase) “if you give enough monkeys typewriters they will write Hamlet.” That could take some time and Google seems more interested in having its computers write something that could be deemed a “lost Shakespeare play.”

It’s not terribly different than what Microsoft is trying to do with its own machine learning technology. Through a collaboration with the Dutch museum Mauritshuis, Delft University of Technology and ING, Microsoft launched Next Rembrandt in the hopes that machine learning and a 3D printer can create a Rembrandt portrait.

Asking artists, coders and machine learning experts to come together, Google wants to see music through algorithms.

“It’s not enough just to sample images or sequences from some learned distribution,” explains Eck.

Given the current state of popular music that’s a big ask, given that Siman Cowell and producers like him have gotten rich of learned distribution.

The future of machine learning and the “warnings” over AI

Skynet and the feared future of machine learning that sees the machines taking over, is a long way’s off, but is it? I have the answer, it is. But when computers themselves can starting writing original code, well, that gives pause but the means of manufacturing and mining, smelting and transportation remain a future you needn’t fear.

No one writing about the dangers of the machine run apocalypse are writing anything more than a tome that could be dug up centuries from now and like Nostradamus, they could be considered an oracle. But we already have the screenwriters of the Terminator films and Philip K. Dick and that future will keep the genre of the distant future or science fiction.

Machines “take jobs” from reasonably unskilled laborers or simply laborers that can’t lift a cars door into place all day without strain, risk of injury or mandated enforcement of labor laws or union negotiated contracts.

That was always going to be the case. I’m really quite frankly surprised that Donald Trump hasn’t promised that companies will be forced to pull out all robots under his rule, er presidency, but it’s a long few months from November.

So they have the brawn, but not the brain yet.

As artists, Ok, that’s a bit scary but not really. Someone still has to sell and buy their work. It’s simply an impressive display of human and computer partnership that will ultimately lead to the mapping of the genome.  Oh wait….

Leave a Comment