Dartmouth 1956 workshop coined the term artificial intelligence

Seventy years ago this summer, a small group of researchers gathered at Dartmouth College and, almost as an afterthought, gave a field its name. The Wikipedia entry on the Dartmouth workshop records that John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon organized the meeting for the summer of 1956, and that McCarthy picked the label 'Artificial Intelligence' partly for its neutrality, avoiding a focus on narrow automata theory and the analog feedback focus of cybernetics.

The stakes for a modern reader sit in the wording of the founding proposal, submitted September 2, 1955. It proposed 'a 2-month, 10-man study of artificial intelligence' at Dartmouth and asserted that 'every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.' That conjecture, hedged and audacious in the same sentence, is the same one every current foundation-model roadmap is still testing.

According to the article, the workshop itself ran roughly six to eight weeks, drew about 20 documented participants, and was pursued after McCarthy approached the Rockefeller Foundation in early 1955 to request funding for a summer seminar for about 10 people. McCarthy's distribution list identified 47 people interested in the topic. Ray Solomonoff, Minsky, and McCarthy stayed for the full duration; Shannon attended for about four weeks; Selfridge, Newell, Simon, and others rotated through. The sessions were mostly brainstorming on the top floor of the math department, and Wikipedia credits the meeting with encouraging the rise of symbolic methods, early expert systems, and the split between deductive and inductive approaches. It has been called 'the Constitutional Convention of AI.'

The honest caveat is that this is a Wikipedia encyclopaedia summary, not new reporting, and it flags that Dartmouth was not the first conference on thinking machines; a 1951 Paris cybernetics conference and other earlier meetings preceded it. The article does not quantify the Rockefeller ask, does not itemize concrete technical outputs from the six weeks, and does not offer a critical assessment of what the workshop got wrong. Read it as a founding-myth document that got the naming right, not as a technical audit.

What is useful today is the reminder that a field's language and self-conception get set at moments like this, often by a tiny group with modest funding. If your board is asking why AI conversations still lean on symbolic framings, or why 'simulate every feature of intelligence' keeps returning as the target, the vocabulary was fixed in a New Hampshire summer by roughly 20 people and a proposal that a machine could in principle simulate any feature of intelligence.