Abstract

Charles Darwin, father of evolution, and Alan Turing, father of computation, have much in common. While much fanfare has been made of the “Turing Test” in recent media, it is less known that in the very same paper wherein Turing described his famous imitation game, he also proposed a “learning machine” which would parallel the principles of evolution. The modern interpretation of this idea, “evolutionary computing”, has found applications from robotic control and circuit design, to places as unlikely as police sketches.

But Darwin’s legacy runs deeper. Turing’s paper is considered by many to have given birth to the field of artificial intelligence, which today struggles with a key question: is there a single algorithm which can learn anything, i.e., a “master algorithm”? This question bears an uncanny resemblance to the one faced by natural philosophers in Darwin’s day: is there a “master algorithm” which explains the origin of species? Darwin’s stunningly elegant answer had an immeasurable impact on scientific progress; the impact of a master algorithm for machine learning remains to be seen.