Early Computers Play Tic-Tac-Toe, Nim, Chess and Draughts

Popularized in today's imagination by DeepMind's AlphaGo, machine intelligence and games has a long history. In fact, in the early 1950s, very early computers around the world started playing basic games, like tic-tac-toe with real people.

First was the 1950 Canadian National Exhibition in Toronto, where Josef Kates’s Bertie the Brain played tic-tac-toe against visitors. Then in 1951, Ferranti’s Nimrod played nim at the Festival of Britain’s Science Exhibition in London. Both machines were built for that one specific game only, though.

In contrast, Dietrich Prinz used the general-purpose Ferranti Mark I at the University of Manchester to solve simple mate-in-two chess problems in November 1951. Finally, between 1951 and 1952 Christopher Strachey (of love-letter fame) wrote a draughts program that looked a few moves ahead and valued positions by the pieces left on the board. Given wrong input, it printed “I REFUSE TO WASTE ANY MORE TIME. GO AND PLAY WITH A HUMAN BEING.”

We have always welcomed the challenge from machines, particularly as they wobbled clumsily in their early days. Now, in the second quarter of the twenty-first century, artificial intelligence can beat humans at essentially any digital task or game. The test has turned to robotics as the next human vs machine game frontier.