The Dartmouth Summer Research Project on Artificial Intelligence, held from June 18 to August 17, 1956, is widely considered to be the event that started the development of AI as a research discipline. Hosted by John McCarthy, Marvin Minsky, Claude Shannon, and Nathaniel Rochester, it brought together dozens of leading thinkers in AI, computer science, and information theory to chart future research paths.

Group photo [shown above] captured the seven main participants. When the photo was reprinted in Eliza Strickland’s October 2021 article “The Troubled Past and Uncertain Future of Artificial Intelligence”, in IEEE Spectrum, the signature lists six people plus one “unknown”. So who was this stranger?

Who is on the photo?

The six people in the photo are easy to identify. In the back row, from left to right, we see Oliver Selfridge, Nathaniel Rochester, Marvin Minsky and John McCarthy. Ray Solomonov sits in front on the left, and Claude Shannon on the right. All six have contributed to artificial intelligence, computer science or related fields in the decades since the Dartmouth workshop.

A close-up black and white photograph of seven smiling men sitting on a lawn.Back row, left to right, Oliver Selfridge, Nathaniel Rochester, Marvin Minsky and John McCarthy. Front left Ray Solomonov; right Claude Shannon. The identity of the man between Solomonov and Shannon remained a mystery for some time.Minsky family

Between Solomonov and Shannon is an unknown person. Years later, some people suggested that it was Trenchard More, another artificial intelligence expert who attended the seminar.

I first came across a group photo of Dartmouth in 2018 while researching materials for Ray’s memorial website. Ray and I met in 1969 and married in 1989; he passed away at the end of 2009. Over the years, I’ve attended a number of his talks and met many of Ray’s colleagues and colleagues in AI, so I was curious about this photo.

I thought, “Gee, that guy in the middle doesn’t look like what I remember about Trenchard.” So I called Trenchard’s son Paul More. He assured me that the stranger was not his father.

More recently, among Ray’s papers, I discovered a letter. On November 8, 1956, Nat Rochester sent a short note and a copy of the photograph to his colleagues: “I enclose a printout of a photograph I took of the artificial intelligence group.” He sent his note to McCarthy, Minsky, Selfridge, Shannon, Solomonov, and Peter Milner.

A typewritten letter with a photograph of six men addressed to six names from Nathaniel Rochester.A few months after the workshop, Nathaniel Rochester sent a copy of the photo, along with this note, to six people.Grace Solomonoff

So the unknown must be Milner! It makes sense. Milner worked on neuropsychology at McGill University in Montreal, although he is an electrical engineer by training. He is not usually confused with other AI pioneers because his research interests are at odds with theirs. Even at Dartmouth he felt he was above his head, as he wrote in his 1999 autobiography: “I was invited to a meeting of computer scientists and information theorists at Dartmouth College…. Most of the time I had no idea what they were talking about.”

In his fascinating autobiography, Milner writes about his work in radar development during World War II and his switch from nuclear reactor design to psychology after the war. His 1954 doctoral dissertation, “The Effects of Intracranial Stimulation on Behavior in Rats”, studied the effects of electrical stimulation on certain rat neurons, which became commonly known as “pleasure centers”.

This work led to one of Milner’s most famous papers, The Assembly of Mark II Cells, in 1957. The article describes how when a neuron in the brain fires, it fires similar connected neurons (especially those already fired by sensory input) and randomly fires other cortical neurons. Cells can form assemblies and connect with other assemblies. But neurons don’t seem to exhibit the avalanche-like behavior of atoms that leads to an exponential explosion. How neurons can suppress this effect was one of his ideas that led to new discoveries at the seminar.

Milner’s work contributed to the early development of artificial neural networks, so he was invited to a meeting in Dartmouth. AI researchers have shown considerable interest in studying the brain and neurons in order to replicate their function and intelligence.

But, as Strickland points out in his October 2021 Range article, a division was already forming in AI research. One side focused on replicating the brain, while the other side was more interested in what the mind could do to directly solve problems. Scholars interested in this latter approach were also represented at Dartmouth and later advocated the development of symbolic logic using heuristic and algorithmic processes, which I will discuss later.

Where was this photo taken?

A 1956 photograph of Rochester shows the left side of Dartmouth Hall in the background. In 2006, Dartmouth convened the AI@50 conference to celebrate the 50th anniversary of the AI ​​gathering and discuss the present and future of AI. Trenchard More, the person most often erroneously referred to as “the unknown” in Nat’s photo, met with the organizers, James Moore and Carey Heckman, as well as Wendy Conquest, who was working on a film about artificial intelligence for the conference. None of the AI@50 organizers knew exactly where the 1956 meeting took place.

Others led them across the lawn to the left side door of Dartmouth Hall. He showed them the rooms that were in use, which in turn brought back old memories. As Mohr recalled in a 2011 interview, during a meeting in 1956, “Selfridge, Minsky, McCarthy and Ray Solomon and I gathered around the dictionary on the booth to look up the word ‘heuristic’ because we thought it might be a useful word. “. During that tour of Dartmouth Hall in 2006, he was delighted to find that the dictionary was still there.

The word “heuristic” was used throughout the summer of 1956. Instead of trying to analyze the brain to develop machine intelligence, some participants focused on the operational steps needed to solve a given problem, especially using heuristics to quickly determine the steps.

For example, at the beginning of the summer, Herb Simon and Alan Newell gave a talk on the program they had written, the Logic Theory Machine. The program was based on early ideas of symbolic logic with algorithmic steps and heuristics in the form of a list. They later received the 1975 Turing Award for these ideas. Think of heuristics as intuitive guides. The logic theory machine used such guides to initiate algorithmic steps, that is, a set of instructions for actually executing a solution to a problem.

Who was not in the photo

There was one person who attended the Dartmouth Seminar from time to time but was never on any of the attendee lists: Gloria Minsky, Marvin’s wife.

But that summer, Gloria was definitely present. Marvin, Ray, and John McCarthy were the only three participants who stayed for the entire eight-week workshop. Everyone else came and went as their schedule allowed. At the time, Gloria was working as a pediatrician at the Children’s Hospital in Boston, but whenever she had the opportunity, she came to Dartmouth, stayed at Marvin’s apartment, and visited everyone who was at the seminar.

A few years earlier, in the spring of 1952, Gloria was doing her residency in pathology at New York’s Bellevue Hospital when she began dating Marvin. Marvin was a Ph.D. a Princeton student, as was McCarthy, and both were invited to Bell Labs for the summer to work under Claude Shannon. In July, just four months after they first met, Gloria and Marvin got married. Although Marvin worked non-stop for Shannon, Shannon insisted that he and Gloria go on their honeymoon to New Mexico.

Letter from John McCarthy to Ray Solomon on Dartmouth College stationery.In March 1956, John McCarthy, one of the organizers of the Dartmouth Artificial Intelligence Seminar, invited Ray Solomonoff to a summer seminar in Hanover, New Hampshire.Grace Solomonoff

Four years later, McCarthy, Shannon, and Minsky organized a seminar in Dartmouth with Nat Rochester. Gloria recalled a conversation between her husband and Ray, in which Marvin expressed the idea that later became one of his hallmarks: “To understand something, you need to see something in more than one way.” In Minsky’s 2007 book Emotion machine, he considered how emotions, intuition and feelings create different descriptions and provide different perspectives on things. He favored symbolic logic and deductive methods in AI, which he called “good old AI”.

Ray, meanwhile, focused on probabilities—the likelihood of something happening and predictions of how it might evolve. He later developed algorithmic probability, an early version of algorithmic information theory, where every single description of something leads to a probabilistic probability (some more likely, some less likely) of a given outcome in the future. Probabilistic methods eventually became the basis of machine learning.

In this day and age with chatbots in the spotlight and compression techniques increasingly being used in AI, the value of understanding things in many ways and using probabilistic predictions will only increase. That is, the logical and probabilistic methods are combined. This, in turn, is helped by new work on neural networks, as well as symbolic logic. So, the photo that Nat Rochester took didn’t just capture a moment in time for the AI. It also gave a glimpse into how AI would evolve.

The author thanks Gloria Minsky, Margaret Minsky,Nicholas Rochester, Julie Sussman, Gerald Jay Sussman and Paul More for their help and patience.

From articles on your site

Related articles online