Learn Like People Do, Silly

Decades ago the AI community kicked around the idea that there was a rationale for teaching machines to learn in ways that resemble how people learn. At the time, there was little appreciation for how difficult that might be, but it was a healthy debate that has been lost in the "I can code that in ROR" wave.

Courtesy of Awesom-O, I was reminded of this by a recent announcement by Anybot that its robot learned to walk using machine learning methods.

No comments: