The June, 2016 issue of Wired magazine includes “The Rise of Artificial Intelligence and the End of Code: what that means for us.” This fascinating article describes recent progress with developing artificial intelligence programs, as illustrated by a recent contest between the world champion for the game Go and a computer program developed by a Google team.
Rather than using traditional computer programming, the Google team said data into a “deep neural network,” a network of hardware and software that roughly mirrors the web of neurons in the human brain. The result is a program that, amazingly, learns from playing games and revises its internal algorithms in response to feedback from game outcomes. The result was a program that showed a version of “in human intuition” that defeated the Go Grand Master in a series of dramatic games. The article acknowledges that this has potentially alarming implications for human-computer “relationships,” but also goes on to describe another remarkable outcome: The Grand Master learned from the program’s strategies, and became a better Go player! The Grand Master is reported to have not lost a match since playing the computer program.
The article concludes, “the machine did a very human thing even better than human. But in the process it made those humans better at what they do.” The author predicts that the future of computing will be, rather than traditional programming, more of a “parent-child training relationship” between humans and computers.