> Samuel's Checkers_

In 1952, Arthur Samuel, a veteran electrical engineer at IBM, sat down at the console of an IBM 701 — one of the first commercially available scientific computers, a machine the size of a room that could perform 17,000 instructions per second — and began teaching it to play checkers. Samuel was not a young hotshot; he was 51 years old, a quiet, methodical man who had spent years working on vacuum tube technology. But he had an idea that was decades ahead of its time: instead of programming the computer with explicit rules about how to play checkers, he would let the machine learn from experience.

Samuel's approach was revolutionary in its simplicity. He programmed the IBM 701 to play games against itself — thousands upon thousands of games, running through the nights and weekends when the expensive machine was not being used for "serious" work. After each game, the program would adjust its evaluation function, strengthening the weights of features that correlated with winning and weakening those that correlated with losing. This self-play mechanism — a machine improving by competing against itself — would not be fully appreciated for another 65 years, when DeepMind's AlphaGo Zero used the same fundamental principle to master the game of Go.

By 1956, Samuel's program could play a "very good" game of checkers. By the early 1960s, it had surpassed Samuel's own playing ability — a moment of quiet but profound significance. For the first time in history, a human had created something that could outperform its creator at an intellectual task. The program was demonstrated on national television in February 1956, astonishing viewers who watched a giant computer defeat a capable human player. IBM's stock price rose 15 points the next day.

Samuel is also credited with coining the term "Machine Learning" in a 1959 paper, which he defined as the field of study that gives "computers the ability to learn without being explicitly programmed." This phrase, tossed off almost casually in a technical paper about a checkers program, would eventually name an industry worth hundreds of billions of dollars. Samuel continued refining his checkers program until 1967, exploring techniques like alpha-beta pruning and book learning. He never sought fame, and he never started a company. He simply wanted to know if a machine could learn. The answer, resoundingly, was yes.