Every field needs a founding moment, a point researchers can point back to and say, “this is where it started.” For artificial intelligence, that moment is widely considered to be a summer workshop held at Dartmouth College in 1956. It was there that the term “artificial intelligence” was coined, and where a small group of researchers laid out an ambitious, almost audacious vision for what machines might someday be able to do.
The Proposal That Started It All
The Dartmouth workshop traces back to a research proposal written in the summer of 1955 by John McCarthy, a young mathematics professor at Dartmouth College, along with Marvin Minsky of Harvard, Nathaniel Rochester of IBM, and Claude Shannon of Bell Labs, already famous as the founder of information theory. The four submitted their proposal to the Rockefeller Foundation seeking funding for a summer research project, and it was in this document that McCarthy first used the phrase “artificial intelligence” in writing. The proposal confidently stated that the group conjectured that every aspect of learning or any other feature of intelligence could, in principle, be so precisely described that a machine could be made to simulate it. That single sentence would end up framing the entire field’s ambitions for decades to come.
Why McCarthy Chose That Name
McCarthy later explained that he deliberately picked the term “artificial intelligence” to distinguish the new field from existing related areas like cybernetics, which was closely associated with the work of Norbert Wiener and had become tangled up with analog feedback systems and specific control-theory ideas McCarthy wanted to avoid. He wanted a neutral name that did not commit researchers to any particular method, whether that meant symbolic logic, neural networks, or something else entirely, leaving room for many different technical approaches to be explored under one banner. That naming decision proved consequential; six decades later, the term is still the default way the world refers to machine intelligence research, even as the underlying techniques have changed almost beyond recognition.
What Actually Happened That Summer
The workshop itself ran for roughly six to eight weeks during the summer of 1956 on the Dartmouth campus in Hanover, New Hampshire. Unlike a modern conference with a packed schedule, it was a loosely structured, informal gathering: attendees came and went throughout the summer rather than all being present the entire time, and there was no fixed agenda of talks or papers. Participants included the four original proposal authors as well as other researchers who would go on to shape the field, including Allen Newell and Herbert Simon, who arrived having already built a working program called the Logic Theorist, arguably the first true AI program, capable of proving mathematical theorems from Whitehead and Russell’s Principia Mathematica. Arthur Samuel, known for his pioneering work on a self-improving checkers-playing program, and Ray Solomonoff, who later made foundational contributions to algorithmic probability, were also present. Attendance fluctuated, funding from the Rockefeller Foundation was smaller than originally requested, and by most accounts the group did not reach any unified consensus on methods or definitions. Despite that lack of resolution, the event succeeded in bringing together, for likely the first time, most of the researchers who would go on to define the field’s early decades.
The Optimism, and the Overconfidence
The Dartmouth workshop is remembered almost as much for its optimism as for its naming contribution. Attendees genuinely believed significant progress on machine intelligence could be made within a matter of months of concentrated effort, not decades. Herbert Simon later predicted in 1957 that within ten years a computer would be the world chess champion, a milestone that would not actually be reached until IBM’s Deep Blue defeated Garry Kasparov in 1997, forty years later. This pattern, bold short-term predictions followed by results that took far longer than expected, would repeat itself throughout AI’s history, eventually contributing to periods of disillusionment and funding cuts in the 1970s and again in the late 1980s that researchers later nicknamed “AI winters.”
The Diverging Paths That Followed
Rather than settling on one unified approach, the Dartmouth workshop effectively planted the seeds for several different research traditions that would compete and coexist for decades. McCarthy went on to develop the Lisp programming language in 1958 and championed symbolic, logic-based approaches to AI for most of his career. Minsky also pursued symbolic reasoning and later became a vocal critic of the simple neural network models popular at the time, co-authoring a 1969 book that is often credited with contributing to reduced funding for neural network research for over a decade. Meanwhile, the neural network tradition that Minsky was skeptical of continued developing in parallel, eventually leading, through the work of later researchers like Geoffrey Hinton, Yann LeCun, and Yoshua Bengio, to the deep learning techniques that dominate AI today. In that sense, the Dartmouth workshop did not just coin a name; it opened a field broad enough to contain approaches that would spend the next several decades alternately competing for dominance.
Why Dartmouth Still Matters Today
Looking back nearly seven decades later, the Dartmouth workshop’s lasting significance lies less in any specific technical result and more in what it represented: the formal recognition of artificial intelligence as a distinct field of study worth pursuing on its own terms, with its own name and its own community of researchers. The questions the Dartmouth attendees asked, about learning, reasoning, language, and abstraction, remain central to AI research today, even though the tools used to pursue them, from symbolic logic to deep neural networks to today’s large language models, have changed dramatically. Every time a new AI breakthrough makes headlines, it is, in some sense, still answering the ambitious and somewhat premature question that a handful of researchers gathered in Hanover, New Hampshire first asked out loud in the summer of 1956.