Arthur Samuel

Arthur Samuel

An American engineer at IBM who began writing a checkers-playing program in 1952 that improved by playing against itself, tuning the weights of its board-evaluation function and remembering positions it had already scored. His 1959 paper on the work introduced the phrase "machine learning", defining it as giving computers the ability to learn without being explicitly programmed. The program was the first significant piece of software to get better at a task through experience rather than through its author rewriting it, and it prefigured self-play training decades before AlphaGo. (See also: Machine learning, RLHF, Demis Hassabis)