What would it take to replicate all this in a piece of software? This is a major challenge in the field of computer science, and one whose full solution is quite some distance away. Indeed, the British mathematician and computer scientist Alan Turing famously posed a version of this challenge in 1950, proposing a useful operational definition for machine intelligence. His proposal, known today as the “Turing Test”, imagines (in contemporary terms) that you are sitting
in front of two instant-messenger windows; in one you can talk to a human being, and in the other you can talk to a computer. If you can’t tell which is which, then the computer can be considered, for all practical purposes, “intelligent”.

It is fascinating, given all the advances in both the science and technology of computing over the past half-century, how prescientTuring was in formulating this challenge: that despite the ability of current computing systems to handle trillions of bytes of data, billions of user queries per day, and computations on massive scales, we still lack the techniques—in automating the processing of human language, synthesizing disparate pieces of knowledge, and drawing
inferences—to replicate the seemingly simple act of keeping up one’s end of a conversation. But if we were to achieve this in a piece of software, and combine it with the massive information resources and instant recall of a Google or a Yahoo!—what a conversation that would be.