Who Actually Invented "Artificial Intelligence"?

The term artificial intelligence appears in a 1955 funding proposal for a summer workshop at Dartmouth College. The author was John McCarthy, then a young mathematician, and the phrase was a deliberate choice.

The field already existed under other names. Norbert Wiener had made cybernetics famous; others wrote about automata studies. McCarthy wanted a name that was neither, partly because he found cybernetics too focused on analogue feedback and too closely identified with Wiener personally. He also wanted something that sounded like it meant what it said.

The proposal was confident to the point of comedy. It suggested that a two-month, ten-person workshop could make significant progress on making machines use language, form abstractions and solve problems reserved for humans. That gathering — the Dartmouth Summer Research Project on Artificial Intelligence, 1956 — is conventionally treated as the founding event of the field, even though it did not produce the breakthroughs promised.

McCarthy later became known for something else: the AI effect. His observation was that once something works, people stop calling it artificial intelligence. Optical character recognition, chess engines, speech recognition and route planning were all AI until they were merely software.

That is worth noticing now. The phrase gets redefined whenever it succeeds — which means arguments about whether something "is really AI" have been running since before the term was a year old.