Abstract

The idea that computation has its own laws and limitations emerged in the 1930s. Some of the early computing pioneers, most notably Turing and von Neumann, already understood that this idea had far reaching implications beyond technology. It offered a new way of looking at the world, in terms of computational processes. Turing and von Neumann themselves pursued this perspective in such areas as genetics, biological development, cognition and the brain.

There has been much progress in the intervening years in understanding computation. The question that arises for our generation is how to exploit this increasing knowledge to obtain insights into the natural world that cannot be obtained otherwise. This talk will focus on biological evolution approached from this standpoint. The scientific question is to determine the molecular mechanism of biological evolution, to a level of specificity that it can be simulated by computer, and to understand why this mechanism can do the remarkable things that it has done within the time that has been available. We argue that the tools needed to approach this come from machine learning, the field that studies how mechanisms that achieve complex functionality can arise by a process of adaptation rather than design.