Google Magenta Creates Blah Music But Proves Technology
posted Sunday Jun 5, 2016 by Scott Ertz
Artificial Intelligence is a computing concept that has been an obsession for many for decades. Science fiction stories are littered with AI, both positive and negative. In the real world, however, AI and neural networks have only recently become a big topic of research because of the power of cloud computing. Distributed computing has been used for long-running computing cycles for years, but now, thanks to the cloud, it can be used for large, short-term processes.
Different companies with distributed computing platforms and neural networks have used it for distinctly different purposes. While Microsoft is using Azure technology to build Cortana, Microsoft Health and the like, IBM has used Watson to design a barbeque sauce. Both have their purposes - Microsoft is building products to show what Azure can do, while IBM is showing the breadth of capabilities of Watson.
Google seems to be following in IBM's footsteps, allowing Magenta to play board games and, as of this week, compose music. The music, available here, isn't very good, but that isn't the point. The point is that the music, minus the drums, was written by first listening to other music, discovering patterns, and then creating something based on it. The music itself will never win an award, but it could help inspire others, like in the linked MP3. The drums were added later, making the music itself sound more engaging.
Douglas Eck, a Google research scientist, said,
Magenta has two goals. First, it's a research project to advance the state of the art in machine intelligence for music and art generation. Machine learning has already been used extensively to understand content, as in speech recognition or translation. 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.
As research projects like this increase, the overall knowledge about how neural networks learn the best, and the best ways to build them. Over time, these networks can be used to increase medical knowledge and much more. This is simply the first step of many, but at the current rate of development, those steps will continue to get faster and we can begin to use the technology to accomplish real good.
