Dartmouth workshop
/ DART-muth WURK-shop /
The Dartmouth workshop was a small summer gathering of researchers at Dartmouth College in the United States in 1956, now remembered as the event that founded artificial intelligence as a named field. It is where the very term 'artificial intelligence' was coined — chosen by the organizer John McCarthy partly to mark out new territory distinct from existing areas like cybernetics. A handful of people meeting for a few weeks gave a scattered set of ideas a shared name and a sense of common purpose.
The 1955 proposal that launched it captured the founders' optimism in a now-famous line: that the study would proceed 'on the basis of the conjecture 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.' Attendees included figures who would shape the field for decades — John McCarthy, Marvin Minsky, Claude Shannon, and others — and the topics they set out, from language and reasoning to learning and neural nets, mapped much of AI's agenda for years to come.
Its significance is more symbolic than technical: no single breakthrough was achieved that summer, and the bold timeline implied by that founding conjecture proved wildly optimistic — the hard parts took far longer than anyone there imagined, and contributed to later disappointments. But the workshop matters because it crystallized a community, a name, and an ambition. When people speak of AI's 'birth' in 1956, this is the moment they mean.
The proposal asked for just two months and ten researchers to make a serious dent in problems like getting machines to use language and improve themselves. Nearly seventy years later, several of those problems are still open — a vivid reminder of how routinely AI's early optimism underestimated the difficulty of the work.
The 1956 proposal: two summer months for problems that are still open today.
Dartmouth gave AI its name and its founding community, not a technical result. The optimism baked into its 1956 proposal — 'a machine can be made to simulate' any feature of intelligence — set a pattern of overpromising that recurred throughout AI's history.