Self correcting algorithms are already beginning to appear. Other than that it is human programmers who work with AI tools to debug them. As neuromorphic computing becomes the standard then machines will learn by observation of their human designers and users. This is where AI is going but we are probably a decade away from seeing neuromorphic technology displacing the current hardware/software duopoly that governs computing today. For some background read: http://www.21stcentech.com/when-will-computers-become-more-human-part-2/, a posting I wrote in 2009. And a more recent piece, http://www.21stcentech.com/artificial-intelligence-image/. You would find George Zarkadakis’ book, “In Our Own Image” a useful way to view the subject.
]]>