After college I went to M.I.T. for my graduate work. I started at M.I.T. in 1959. I had to spend the first year and a half completing coursework for the Ph.D. qualifying exams and taking the exams. With that behind me it was time to choose my M.S. thesis topic. I decided that I wanted my thesis to be in the area of artificial intelligence.
In 1961 Artificial Intelligence was just coming into its own, and M.I.T. was the leader. Two men, Marvin Minsky and John McCarthy, both born in 1927, founded the M.I.T. Computer Science and Artificial Intelligence Laboratory in 1959, the year I entered M.I.T. These two men are the acknowledged leaders in the field. Minsky is still at M.I.T. McCarthy went to Stanford in 1962, where he remained until his death in 2011.
The C.S. & A.I. lab in 1961 was a wild place. Many of the A.I. graduate students had their desks in the same room, and it seemed to me noisy and chaotic. Darts were always being thrown (at a dartboard) as were wadded up pieces of paper (at each other), all of this amidst lively discussions and arguments about A.I. I don’t know how anyone got anything done in that atmosphere, but many of the groundbreaking advances in A.I. happened there.
In 1950 Alan Turing had proposed a test, now called the “Turing test” to determine whether a machine was intelligent. In the Turing test you sit at a teletypewriter and converse with a person out of sight at another teletypewriter, or is it a machine? If it is a machine that can fool you into thinking it is a person, then the machine is intelligent. Unfortunately, Turing introduced the idea by describing a party game in which you are trying to determine if the unseen person is a man or a woman, thus complicating his explanation by introducing the notion of sex which for a while obscured the simplicity of his test.
The common belief in the C.S.& A.I. Lab was that we would achieve true machine intelligence, a machine that could pass the Turing test, probably within five years, certainly within ten years. Many others believed it also. At M.I.T during 1964 – 66 Joseph Weizenbaum wrote a program called ELIZA to analyze natural English sentences. One of the scripts he wrote for ELIZA, DOCTOR, simulated a Rogerian psychotherapist. This was an easy target, since Rogerian therapists typically work by just asking questions. ELIZA was not at all intelligent. Weizenbaum was focusing only on analyzing English sentences. However, many people, including many psychotherapists, saw it as much more and thought that machined could revolutionize the field of psychotherapy.
ELIZA DOCTOR is still available on the internet. It is fun. If it does not understand your sentence it typically replies “Tell me more about your father.” It is hard to believe, today, that anyone could have thought it intelligent.
I’ve often thought about why machine intelligence, which we have yet to achieve, is so much harder than we thought fifty years ago. I think that a central issue is world view. When we talk with anyone else, we depend on the fact that they have a view of the world similar to ours. That is what makes communication across cultures sometimes difficult. Since we share a vast amount of common information with other people, we communicate in shorthand, taking for granted that shared information. Computers lack that comprehensive world view.
One of the most impressive recent advances in A.I. is the IBM program Watson, which recently won the game of Jeopardy playing against two human experts. Jeopardy is a very interesting challenge because it uses colloquial English, but more importantly it depends on having an excellent world view. Watson got it’s world view by mining the internet. So the world view shared by all humans is now available to machines. How far are we now from true machine intelligence?
– Rudd Canaday (ruddcanaday.com)
– Start of blog: My adventures in software