Best AI Applications of the Week—December 28, 2017
AI-Powered Prosthetic Allows Musician to Play the Piano Again
Playing the piano requires complex and dexterous finger movements, controlled by the musician’s brain. After losing his arm, musician Jason Barnes thought he would never play two-handed again—until researchers at Georgia Tech developed a prosthetic arm that allowed him to do so.
It uses a deep neural network trained to sense, through ultrasound, the muscle movements in the arm and predict which finger the musician wants to move.
Where Do I Get a Shirt Like That?
You’ve spied a shirt you really like, but it’s more than you’re interested in spending on a piece of clothing. Is there an easy way to find it for less? Not yet, but soon. In early 2018, Gauss, a Japanese AI company in the fashion industry, will launch its new tool that takes a picture it’s given and finds similar items for sale. Initially, though, it’ll only be for fashion companies, but soon the company’s aim is to offer you, the consumer, a version.
Watch it at work here.
Is That an AI or a Human Talking? Nobody Knows.
Chalk another one up for deep learning from Google. With the tag team of two deep learning networks—one to translate text into a spectrograph (which represents audio frequencies across time) and another to generate speech from that spectrograph—Google can generate voice that is purportedly indistinguishable from a human. If you don’t think you can be fooled, judge for yourself.
Unique of the Week
If you’re worried about autonomous killer robots, you may not enjoy knowing that AI can teach itself how to walk, without any instruction from researchers on what it takes to do it. Whether given a human, spider, or stick-like form, Google’s DeepMind successfully taught itself how to move from point A to point B—but if killer robots taught themselves to walk like this humanoid form, it’d most definitely be one of your worst nightmares. Watch it at work here.