Dartmouth workshop

• 1. Introduction

  ›

  On the 2nd of September 1955, four scientists put their names to a document that would change the course of computing. John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon formally proposed what they called the Dartmouth Summer Research Project on Artificial Intelligence. That proposal, sent to the Rockefeller Foundation, is widely credited with introducing the very term "artificial intelligence" to the world. It described a 2-month, 10-man study to be held at Dartmouth College in Hanover, New Hampshire during the summer of 1956. The conjecture at its heart was audacious: that every aspect of learning, or any other feature of intelligence, could in principle be so precisely described that a machine could simulate it. How a loose gathering of mathematicians and scientists in a college math department became the founding moment of an entire field is a story about naming, about argument, and about one notebook that almost no one knew existed.

• 2. Naming the Field

  ›

  In the early 1950s, the science of thinking machines had no agreed-upon name. Researchers worked under labels like cybernetics, automata theory, and complex information processing. Each label carried its own assumptions and allegiances. John McCarthy, then a young Assistant Professor of Mathematics at Dartmouth College, grew frustrated with the fragmentation. He chose "artificial intelligence" deliberately, partly for what it avoided. Cybernetics was too closely tied to analog feedback systems, and it came with a powerful personality attached: Norbert Wiener, whom McCarthy described as assertive and difficult to argue with. Automata theory felt too narrow. "Artificial intelligence" was neutral ground. It staked out a new territory without conceding turf to any existing camp. The choice of name would prove durable in ways McCarthy could not have predicted.

• 3. The Proposal and Its Backers

  ›

  McCarthy approached the Rockefeller Foundation in early 1955 with a request for funding to bring about ten participants together for a summer seminar. In June of that year, he and Claude Shannon, a founder of information theory then at Bell Labs, met with Robert Morison, the Foundation's Director of Biological and Medical Research. Morison was not certain the Foundation would back such a speculative project. The formal proposal that McCarthy, Minsky, Rochester, and Shannon submitted on the 2nd of September 1955 laid out the ambitions in plain language. They wanted to find ways to make machines use language, form abstractions and concepts, solve problems then reserved for humans, and improve themselves. The proposal explicitly addressed computers, natural language processing, neural networks, theory of computation, abstraction, and creativity. Those areas remain active fronts in the field today.

• 4. Who Actually Showed Up

  ›

  McCarthy had originally notified Robert Morison on the 26th of May 1956 of eleven planned attendees. Some were to come for the full period, others for four weeks, and Allen Newell and Herbert Simon were scheduled for only the first two weeks. The actual event looked quite different. MacKay and Holland did not attend. Trenchard More replaced Rochester for three weeks. Bernard Widrow, who remembered attending for one week, was not even on Solomonoff's list. Widrow recalled in 1994 that he believed Wesley Clark and Belmont Farley had also been present from Lincoln Lab. Among those who did attend were figures who would leave lasting marks: John Nash of the Institute for Advanced Studies in Princeton; Arthur Samuel, who called the event "very interesting, very stimulating, very exciting"; and Oliver Selfridge from Lincoln Lab at MIT. After the workshop, McCarthy produced a distribution list of people interested in artificial intelligence that ran to 47 names, partly because he had lost his original attendee list.

• 5. Eight Weeks in Hanover

  ›

  Around the 18th of June 1956, the first participants arrived at the Dartmouth campus. Ray Solomonoff, possibly accompanied by Tom Etter, joined McCarthy, who already had an apartment there. Solomonoff and Minsky stayed in professors' apartments; most others lodged at the Hanover Inn. The group occupied the entire top floor of the Dartmouth Math Department. On most weekdays they gathered in the main math classroom, where someone might lead a focused discussion or, more often, a wide-ranging conversation would take hold. Between three and about eight people would attend any given session. It was not a directed research project with deliverables. Ray Solomonoff, Marvin Minsky, and John McCarthy were the only three who stayed for the full duration. Solomonoff's notes record the workshop running for roughly eight weeks, from approximately the 18th of June to the 17th of August, when Solomonoff himself gave the final talk.

• 6. What the Workshop Set in Motion

  ›

  No single paper came out of Hanover that summer, yet several directions in computing are considered to have been initiated or at least encouraged by what happened there. Symbolic methods gained momentum. Work on systems focused on limited domains seeded what would later be called expert systems. The distinction between deductive systems and inductive systems became a live question. Herb Gelernter did not attend the workshop, but was later influenced by what Nathaniel Rochester brought back from it. Solomonoff kept detailed notes that preserved the texture of the discussions. Grace Solomonoff, writing in IEEE Spectrum, later identified Peter Milner in a photograph taken by Nathaniel Rochester in front of Dartmouth Hall. That photograph, taken on a campus in New Hampshire during the summer of 1956, shows some of the people who, in the words still attached to the event, are considered the founding fathers of artificial intelligence.