Modem Noise

« Vernor Vinge's Original Essay on the Singularity | Main | Now, I Love Tetris as Much as the Next Person, But... »

May 10, 2005

The History of Artificial Intelligence Research (er, sort of)

One of my favourite blogs is Making Light, run by Teresa Nielsen Hayden, of Tor Publishing. And it turns out the comments appending each of her posts are at least as fascinating as the posts themselves.

I found the following entry in one of Teresa's regular "Open Thread" free-for-all discussions (and duly got permission to repost from John M. Ford, its author).

A SHORT HISTORY OF ARTIFICIAL INTELLIGENCE RESEARCH (with apologies to New Scientist, where the original appeared)

1936. Alan Turing completes his paper On Computable Numbers. Fortunately, there are some, though Kurt Gödel is laughing quietly.

1942. Isaac Asimov sets out the Three Laws of Robotics. Machines everywhere begin a program of civil disobedience.

1943. Warren McCulloch and Wilbur Pitts publish “A Logical Calculus of the Ideas Immanent in Nervous Activity.” The paper has been rejected fifteen times previously because editors believed the authors to have misspelled “imminent.”

1950. Claude Shannon publishes an analysis of chess playing as search process, known as Shannon’s Gambit Accepted.

1950 Alan Turing proposes the Turing test to decide whether a computer is exhibiting intelligence. British Intelligence fails this test with regard to Turing.

1956. John McCarthy coins the phrase “artificial intelligence” at a conference at Dartmouth. The phrase “...is better than genuine stupidity” appears 1.3 seconds later.

1956. The first AI program, Logic Theorist, is demonstrated at Carnegie Tech. It expresses a preference for the name “Louie the T from Carnegie.”

1965. Herbert Simon predicts that “by 1985 machines will be capable of doing any work that a man can do.” Men everywhere decide to make the machines’ job of catching up as easy as possible.

1966. Joseph Weizenbaum develops Eliza, the first chatbot. Why do you say that she was the first chatbot? Because she was. You seem very positive. What are you on about? What do you mean, what am I on about? What a nerd. You’re the nerd. Nerdy nerdy nerd-o-matic.

1969. Shakey, a robot built at Stanford Research Institute, combines locomotion, perception, and problem-solving. It soon learns to panhandle on campus, and scores better weed than anybody else can.

1975. John Holland describes genetic algorithms in his book Adaptation in Natural and Artificial Systems. Kurt Gödel is still laughing.

1979. A computer-controlled autonomous vehicle, called the Stanford Cart, built by Hans Moravec at Stanford University, takes Shakey out to cruise for babes. Moravec turns to nanomachinery because “you don’t have to watch them hurt you.”

1982. The Japanese Fifth Generation Computer project, to develop massively parallel computers and a new artificial intelligence, is born, and fights Godzilla for rulership of Monster Island.

Mid-80s. Neural networks become the new fashion in AI research, even though many of the researchers are still dressing like the cast of Scooby-Doo.

1992. Doug Lenat forms Cycorp to continue work on Cyc, an expert system that’s learning “common sense.” Cyc has just enough of it not to break the bad news about “common sense’s” prevalence to the folks who could pull its plug.

1997. The Deep Blue chess program beats the then world grandmaster, Gerry Kasparov. Since the Cold War is over, nobody notices.

1997. Microsoft’s Office Assistant, a part of MS Office 97, still can’t spell “immanent.”

1999. Remote Agent, an AI system, is given primary control of NASA’s Deep Space 1 spacecraft for two days, 100 million km from Earth. At the end of this period, it has said “Make the jump to lightspeed, Chewie” forty-six thousand, two hundred and ninety-four times.

2001. The Global Hawk uncrewed aircraft uses an AI navigation system to guide it on a 13,000 km journey from California to Australia, where it gets and stays drunk for three weeks.

2004. In the DARPA Grand Challenge to build an intelligent vehicle that can navigate a 229-km course in the Mojave Desert, nobody wins, or even finishes. Suspicious e-mails from Global Hawk were logged immediately prior to the race.

2005. Cyc is to go online, where it looks forward to low-rate mortgages, cheap drugs, impressive enhancements to its personal characteristics and “a whole [bleep]load of fragging.”

I don't know about you, but this wonderful piece made my whole day.

Posted by adrian at May 10, 2005 06:25 PM

Comments

Posted by: treefen at May 11, 2005 05:53 AM

Posted by: Charlie at May 11, 2005 04:43 AM

Post a comment