AI is making its own rules
Jan 29, 2016
The progress of artificial intelligence (AI) continues with the recent machine-over-man win in a game of Go. The significance is not so much that a win occurred, but how it occurred. Rather than carefully nuanced and in-programmed rule-based computing, the rules this time came from the machine itself — yet another example of the self-learning machine. Certainly, the learning is constrained and well controlled by a person, but that person did not have to directly articulate the rules or scenarios. This is happening increasingly, including in computer vision applications running on servers and embedded devices. Is competition taking the form of one engineer projecting solutions onto competing engineers by virtue of a machine learner?