Checkers at the origins of AI and Machine Learning


Share on linkedin
Share on twitter
As Machine Learning is becoming more and more popular and concrete applications are observed in our daily lives, I find it useful to rewind at the roots of AI and ML, back in 1956. And to my big surprise, I learned that checkers, not check was at those origins.

With this article, my intention is to look back to history and to go to the origins of Artificial Intelligence and Machine Learning. 

Arthur Samuel playing checkers on the IBM 701
Arthur Samuel playing checkers on the IBM 701 was demonstrated to the public on television on February 24, 1956

February 1956: A demo of a computer playing checkers

The first checker program for the IBM 701 mainframe was written in 1952 but the first program with learning was completed in 1955.  Arthur Samuel demonstrated it on TV on February 24, 1956.

Invitation to The Dartmouth Summer Research Project on Artificial Intelligence, finally hold in June 1956

June 1956: The first usage of "Artificial Intelligence"

John McCarthy, a young (29 y-o) Assistant Professor of Mathematics at Dartmouth College proposed a conference in the summer of 1956 with 20 attendees, including Arthur Samuel. This is known as the first time the term “Artificial Intelligence” was coined. 

We propose that a 2 month, 10 man study of artificial intelligence be carried out during the summer of 1956 at Dartmouth College in Hanover, New Hampshire. The study is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.

John McCarthy
Some Studies in Machine Learning Using the Game of Checkers, Arthur Samuel, 3, July, 1959

July 1959: The first usage of "Machine Learning"

On July 1959, Arthur Samuel is publishing “Some Studies in Machine Learning Using the Game of Checkers“, the first known document defining “Machine Learning“.  Checkers was used instead of chess as it was relatively more simple.

Two machine-learning procedures have been investigated in some detail using the game of checkers. [...] Enough work has been done to verify the fact that a computer can be programmed so that it will learn to play a better game of checkers than can be played by the person who wrote the program.

Arthur Samuel

With classical programming language, computers are told to do exactly what to do. With Machine Learning, you don’t give detailed explicit instructions. Instead, you give the “machine” the data and tools it needs to study the problem and solve it without being told what to do. You also give the computer the ability to remember what it did, so that it can adapt, evolve, and learn.

1997: IBM Deep Blue beat Garry Kasparov at chess

Gary Kasparov vs IBM Deep Blue in 1996 and 1997

Much later, the Development of Deep Blue chess software began in 1985 at Carnegie Mellon University and IBM hired quickly the team. In 1996, Deep Blue won 1 game out of 6 against world chess champion Garry Kasparov but Kasparov was still won by Kasparov. The second time, in 1997, Deep Blue won the match.

IBM Deep Blue used brute force computing to execute alpha-beta search (aka Alpha-beta pruning) algorithms in parallel.  In 1997, Deep Blue was the 259th most powerful supercomputer in the world.


Arthur Samuel, Wikipedia

The IBM 700 Series, Computing Comes to Business, IBM

Some studies in machine learning using the game of Checkers, CiteSeerX, Arthur Samuel, 1959

A ‘Brief’ History of Game AI Up To AlphaGo, Andrey Kurenkov, April 18, 2016

Photo credits

Photo by Hassan Pasha on Unsplash

Xl2085, CC BY-SA 4.0, via Wikimedia Commons

Gary Kasparov vs. IBM Deep Blue, Peter Morgan/Reuters

Leave a Reply