ESHS/HSS Annual Meeting

The Play’s the Thing: Games and the Dream of Machine Intelligence in Britain, 1945-1956

• In Event: Alternative Sites of Knowledge Production

English Abstract

In 1997, IBM’s chess-playing computer, Deep Blue, defeated the world champion, Garry Kasparov. The reaction to this event was rapturous, for a machine had “dominated man in the one area where he was unique and unchallenged—his intelligence.” This sentiment reappeared in 2016 after the victory of Google’s AlphaGo over Lee Sedol. But why were chess and Go seen as proving grounds for such lofty claims in the first place? In other words, how and why had games become a test of intelligence, especially for a machine?

In the mid-twentieth century, British computer scientists oversaw games of mechanical chess, built custom machines for tic-tac-toe, and programmed digital computers to play checkers—each the first of their kind. These machines were conceived as experiments in artificial intelligence (AI) and their results were hailed as successful in this sense by the press (and by some historians since). Their designers, however, were less sure.

Utilizing archival materials and personal papers, I highlight how, in reflecting on games and the unintelligent way their machines played them, figures like Alan Turing and Christopher Strachey developed a new theory of AI. As Turing put it, while it was “quite easy” to write a program which would “enable [a machine] to win” at chess, intelligence “consists in a departure from the disciplined behaviour of [this type of] computation.” They wanted to build a machine with the ability “to surprise…to act randomly…to make mistakes,” in other words, to play. In recovering this ambition and the engagement with games that underpinned it, my paper challenges both canonical histories of AI that mythologize the 1956 Dartmouth Research Project as the birthplace for its theories and our contemporary preoccupation with winning games (à la Deep Blue or AlphaGo) as the measure of artificial intelligence.

Author

• Ben Francis, New York University