Room 2 · The Birth · 1952

The Program That Learned to Play Checkers

The Program That Learned to Play Checkers (Arthur Samuel · IBM, 1952): A machine that improves with practice.

From 1952, Samuel developed on IBM computers a checkers program that played against itself and adjusted how it evaluated positions. In time it beat its own author. In 1959 he described that idea with the phrase “machine learning”.

Arthur Samuel joined IBM in 1949 and wrote a checkers program for the company's first commercial scientific computer, the IBM 701. Checkers was a good choice: the rules are simple, but the game is far too large to solve by looking at every possible move.

The program looked a few moves ahead and scored the resulting positions with a formula that weighed features such as the number of pieces, kings and control of the center. The novelty was that the program could change the weights of that formula by itself. It played many games, including games against copies of itself, and kept the adjustments that led to better results. It also remembered positions it had already analyzed, a technique Samuel called rote learning.

In 1959 he described this work in the IBM Journal of Research and Development under the title "Some Studies in Machine Learning Using the Game of Checkers." The paper helped popularize the term "machine learning" and put a clear idea on the table: instead of programming every rule of good play, program a way for the machine to improve from experience.

The program reached a respectable amateur level. In 1962 it won a game against Robert Nealey, a strong player, an event IBM promoted widely. Later programs went much further: in 2007 a team at the University of Alberta proved that perfect checkers ends in a draw.

Samuel kept working on the program for years, at IBM and later at Stanford, where he continued to teach and write into his eighties.

Ask your AI: what is the difference between a program that follows rules and a program that learns its own rules? Ask it to explain using a game you like.

Did you know? As the story goes, a televised demonstration of his program in 1956 made IBM's stock rise.

Source: doi.org/10.1147/rd.33.0210

See it in the museum →

← SNARC, the Rat in the Maze · The Dartmouth Workshop →

All of Room 2: The Birth

Cite this: Nazareno, I. (2026). The Program That Learned to Play Checkers (1952). New Mind Museum. Retrieved November 17, 2026, from https://newmindmuseum.com/museum/the-program-that-learned-to-play-checkers-1952/