In 1966 Joseph Weizenbaum, a professor at the Massachusetts Institute of Technology (MIT), published one of the most celebrated computer programs of all time. The program interacted with a user sitting at an electric typewriter, in English, in the manner of a Rogerian psychotherapist. Weizenbaum called the program "Eliza" because "like the Eliza of Pygmalion fame, it could be taught to 'speak' increasingly well". In his paper, Weizenbaum gave a slightly contrived example of a conversation between Eliza and a young woman (Eliza's responses appear in capital letters):
Men are all alike
IN WHAT WAY?
They're always bugging us about something or other
CAN YOU THINK OF A SPECIFIC EXAMPLE?
Well, my boyfriend made me come here
YOUR BOYFRIEND MADE YOU COME HERE?
He says I'm depressed much of the time
I AM SORRY TO HEAR YOU ARE DEPRESSED
And so on. To Weizenbaum's consternation, some users came to believe that Eliza had real understanding and they would unburden themselves in long computer sessions. Even a few psychotherapists thought Eliza was a potentially useful tool. (If you want to try Eliza, there are lots of reincarnations on the web – type "Eliza" and "program" into Google to find one.)
In fact, the program was something of a trick (albeit a very clever one). All the program did was to decompose the user's input into its constituent parts of speech, and type them back at the user in a manner that sustained the conversation. Eliza was a sensation on the MIT campus and quickly spread to other universities. Weizenbaum was so disturbed that naive users could put their faith in a relatively trivial program, that it changed the future course of his life: he became an advocate for social responsibility in science and a critic of artificial intelligence (AI).
Joseph Weizenbaum was born in Berlin in 1923 to Jechiel Weizenbaum, a furrier, and his wife Henrietta. In 1936 the family emigrated to the United States, where they settled in Detroit, Michigan. Joseph attended the local high school and in 1941 enrolled for mathematics at Wayne University in Detroit. His studies were interrupted in 1941 for war service with the US Army Air Corps as a meteorologist. He resumed his studies after the war, and completed an MS in mathematics in 1950. Wayne University became home to one of the first experimental electronic computers and Weizenbaum stayed on to work with it. One day a local teacher, Ruth Manes, brought her class on a school visit to see the new machine. They subsequently married and had four daughters and a son.
In 1953, Weizenbaum left Wayne University to work in the nascent computer industry on the West Coast. In 1955 he joined General Electric, which was developing a computer called ERMA – a massive bank automation system for the Bank of America – of which he became programming manager. While at General Electric, Weizenbaum developed a novel programming system called SLIP (for Symmetric List Processor), which he published in 1963. It established his reputation and he was offered a teaching position at MIT. Weizenbaum never fully understood how he came to be appointed to one of the world's top computer science departments – he was sure he would never have been accepted as a student there on his merits.
While at MIT, Weizenbaum used SLIP to write the Eliza program. The story of Eliza was among the earliest pieces of AI reportage. AI soon became a hot topic, leading to wild speculation about its potential for the future of mankind, which the AI community itself did little to discourage. Weizenbaum took the view that even if it became possible to build truly intelligent machines (which he doubted) mechanistic reasoning should never be a substitute for human decision-making, which included not just logical deduction but emotional and ethical subtleties that could not be codified.
In 1972, Weizenbaum was granted two years' leave from MIT to write a critique of AI, Computer Power and Human Reason. He consulted, and was encouraged by, leading sociologists and humanists, including Lewis Mumford and Noam Chomsky. Although popular and accessible, Computer Power and Human Reason was a work of rigorous scholarship in computer science, philosophy, and the social sciences. On publication in 1976, it became a bestseller and was translated into several languages. Thereafter Weizenbaum became AI's sternest critic, a technology pundit, and a sought-after public speaker.
Joseph Weizenbaum, computer scientist: born Berlin 8 January 1923; Associate Professor, Massachusetts Institute of Technology 1963-67, Professor 1967-88 (Emeritus); married Ruth Manes (one son, four daughters; marriage dissolved); died Groeben, Germany 5 March 2008.