The Master Algorithm

Pedro Domingos explains how machine learning, the automation of discovery, lets computers program themselves, and describes the race in research labs to build one learner able to discover any knowledge from data. He sorts the field into five schools, connectionism, Bayes' theorem, inductive reasoning, analogical modelling and evolutionary computation, and explains each through brain connections, probability, logic, similarity judgments and natural selection. Each, he suggests, could feed into a unifying master algorithm, and he briefly reviews his own invention, the Markov logic network. Domingos wrote it to interest readers outside the field, and Bill Gates later recommended it as one of two books for understanding AI.

Mark Cuban said it convinced him that the time had come for AI after he had doubted it, and Vinod Khosla suggested it as one view of what a master AI might need.