Arthur Samuel (computer scientist)

Arthur Lee Samuel
Arthur Lee Samuel
Born	5 December 1901 (age 124); Emporia, Kansas, U.S.
Died	29 July 1990 (aged 88)
Occupation	Computer scientist
Known for	Coined the term "machine learning", Samuel Checkers-playing Program, contributions to TeX
Notable work	Samuel Checkers-playing Program, First Grade TeX: A Beginner's TeX Manual

Arthur Lee Samuel (December 5, 1901 – July 29, 1990) was a trailblazer in the fields of computer gaming and artificial intelligence in the United States. He is credited with coining the phrase "machine learning" in 1959. His Samuel Checkers-playing Program was one of the earliest examples of a successful self-improving software, illustrating fundamental AI principles. Additionally, Samuel made significant contributions to the TeX community by authoring an early manual for TeX in 1983.

Biography

Born on December 5, 1901, in Emporia, Kansas, Samuel completed his undergraduate studies at the College of Emporia in 1923. He then obtained a master's degree in Electrical Engineering from MIT in 1926 and served as an instructor there for two years. In 1928, he began working at Bell Laboratories, where his focus was on vacuum tube technology and enhancing radar systems during World War II. Among his inventions was a gas-discharge transmit-receive switch (TR tube) that allowed a single antenna to be used for both transmitting and receiving signals. After the war, Samuel accepted a professorship in Electrical Engineering at the University of Illinois at Urbana–Champaign, where he initiated the ILLIAC computer project but departed prior to its completion.

In 1949, Samuel joined IBM's Poughkeepsie facility, marking the period of his most impactful work. He developed one of the first software hash table implementations and contributed to early IBM research on applying transistor technology in computing. At IBM, Samuel created the first checkers program running on their initial commercial computer, the IBM 701. This program not only demonstrated significant progress in both hardware and software but also caused IBM's stock price to rise by 15 points overnight. His groundbreaking efforts in programming beyond numerical calculations influenced processor instruction set design by broadening computer application scopes. Renowned for his lucid technical writing, Samuel authored an introductory article for one of the earliest computing journals in 1953.

Samuel retired from IBM in 1966 and subsequently joined Stanford University as a faculty member, where he remained active until his passing. At Stanford, he collaborated with Donald Knuth on the TeX typesetting system, contributing to its documentation and continuing software development well into his late eighties.

In recognition of his achievements, he was honored with the Computer Pioneer Award by the IEEE Computer Society in 1987. Samuel died on July 29, 1990, due to complications related to Parkinson's disease.

Computer checkers (draughts) development

Samuel is most renowned within AI circles for his groundbreaking work on computer checkers beginning in 1959 and his foundational studies in machine learning dating back to 1949. After graduating from MIT and teaching there as well as at UIUC between 1946 and 1949, he regarded programming computers to play games as an effective strategy for developing general problem-solving abilities. He selected checkers because it combined strategic complexity with relative simplicity. His program utilized a search tree representing all possible board configurations reachable from the current position. Given memory constraints at the time, he implemented what is now recognized as alpha-beta pruning. Rather than exhaustively exploring all paths to game conclusion, Samuel devised a heuristic evaluation function that scored board states based on criteria such as piece count, number of kings, and proximity to promotion squares. The program employed a minimax algorithm to choose moves that maximized this evaluation under the assumption that the opponent would do likewise.

To enhance performance further, Samuel introduced learning mechanisms into his program. Using rote learning, it stored previously encountered positions along with their associated reward values, effectively increasing search depth through experience accumulation. Subsequent versions refined this reward function by incorporating data derived from professional-level games and improved their strength by engaging in thousands of self-play matches. This methodology enabled Samuel's checkers program to achieve strong amateur-level proficiency—the first board game software to reach such skill—and by the mid-1970s it was capable of competing against skilled amateur players.

Awards

1990. Inaugural Fellow of the Association for the Advancement of Artificial Intelligence
1987. Recipient of the Computer Pioneer Award: Recognized for Adaptive non-numeric processing.

Selected works

Computing bit by bit, or Digital computers made easy (1953). Published in Proceedings of the Institute of Radio Engineers 41, pages 1223-1230.

Groundbreaking contributions to machine learning.

Included with commentary in Computers and Thought, edited by Edward Feigenbaum and Julian Feldman (New York: McGraw-Hill, 1963), pages 71-105.

1983. First Grade TeX: A Beginner's TeX Manual. Stanford Computer Science Report STAN-CS-83-985 (November 1983).

Key figure within the TeX user community.

External links