The use of AI to make modern technology even better and more intuitive is no secret, and given its size and scope, Google is naturally leading the way. The California-based company is now investing in its Brain AI group – a group that has pioneered the use of artificial intelligence in Inbox, Translate, and Photos and is now focused on a new system named Magenta. Unlike any other AI, Magenta will measure how computers can be trained to create music, video or art.
To be launched next month, Magenta was briefly introduced at a music and technology festival in North Carolina named Moogfest recently. Douglas Eck, a researcher at Google's Brain AI Group, demonstrated how Magenta will make use of Google's TensorFlow Open Source engine to find out if machines can be powered using artificial intelligence to create music or paintings. As far as music is concerned, while it is a no-brainer that the initial training process will be long and hard given that there is no involvement of talent or acumen unlike with humans, computers will be fed a number of individual musical notes and will be tasked with joining them and creating a piece of music. The output can be further improved by inserting several variables and emphasizing on creativity. Once Magenta is thrown open to developers next month, developers will be able to feed MIDI music files to TensorFlow and will be able to determine if their computers are able to create new musical patterns or tunes. However, no matter how much training is imparted to computers, the technology isn't yet crisp enough to let them generate the degree of creativity and invention that humans are capable of.
Magenta isn't really a one-of-a-kind platform which will use the full potential of AI to make machines develop independent human-like capabilities. Google has also created DeepDream, a visual AI which works on a similar concept to convert photos into psychedelic art, that is, fusing simple images with those of tall towers, eyeballs or home products. Back in January, Google's DeepMind Group also created an AI named AlphaGo with a singular aim of beating a professional Go player in five games in a row and happily for the company, the AI achieved the task after several attempts. Last week, Google's CEO Sundar Pichai said that Google was looking at creating new AI technologies which would not require physical computing devices to help users with simple tasks. As of now, a number of AI tools are used by Google, like RankBrain to filter search results with a success rate of as much as 80 percent, as well as in its self-driving car which can navigate on their own, find directions and regulate speed as per traffic densities or signals.