How Artificial Intelligence Was Actually First Programmed

By WonderBooks · Aug 20, 2026

The phrase "artificial intelligence" is younger than most people expect — coined for a 1956 summer workshop that ran two months and drew about ten researchers.

Long before the term "artificial intelligence" existed, mathematician Alan Turing had already laid its theoretical groundwork. His 1936 paper on computable numbers described, in the abstract, the kind of machine that could carry out any calculation a set of rules allowed — the concept now known as a Turing machine, and the basis for how every computer since has been understood to work. In 1950, working at the University of Manchester on one of the world's first stored-program computers, Turing proposed a practical test: have a human judge hold a text conversation with both another human and a machine, and see whether the judge could reliably tell which was which. If not, he argued, the machine's performance was worth calling intelligent, whatever was happening underneath it.

The Workshop That Named the Field

The term "artificial intelligence" itself was coined for a specific event. In 1955, four researchers — John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon — proposed a summer study at Dartmouth College, describing it as a "2-month, 10-man study" into whether machines could be made to use language, form abstractions, and solve problems then reserved for people. The workshop ran through the summer of 1956, roughly eight weeks, and while it did not produce a finished technology, it gave the emerging field its name and brought its founding researchers into direct contact with one another for the first time.

The First Program That Looked Like It Was Thinking

The most concrete early result actually predates the Dartmouth workshop by several months. Over the 1955 Christmas holiday, researchers Allen Newell and Herbert Simon completed a program called the Logic Theorist, designed to prove mathematical theorems the way a person would — by searching through logical steps rather than brute-force calculation. It went on to prove a number of theorems from Principia Mathematica, the foundational work of mathematical logic, and is generally regarded as one of the first true AI programs. Newell and Simon followed it with the General Problem Solver a few years later, while John McCarthy — one of the Dartmouth organizers — created the Lisp programming language in 1958, which became the primary language of AI research for decades afterward.

A Field That Kept Overpromising, Then Delivering Late

Early success bred early optimism, and by the mid-1960s, AI research in the United States was drawing significant government funding, with dedicated laboratories opening around the world. That optimism outran the technology available at the time — problems that looked simple turned out to require far more computing power and far better methods than 1960s and 1970s machines could provide, and the field went through two well-documented downturns, commonly called "AI winters," in the 1970s and again in the late 1980s. Revival came in stages: expert systems drove renewed interest in the 1980s, and the growth that has followed since the 1990s has been powered by machine learning, deep learning, and access to far larger datasets than the field's founders had any way to imagine.

Told as an illustrated story