Until
next week, 
Intelligent Machine Post:
The joke's on Sir James?
Dateline: August 2, 1998
Knock Knock . . .
Stanford University's Clifford Nass told the American Association for Artificial Intelligence (AAAI) convention recently that social interaction is a key and often overlooked element of hardware and software design. Machines should be designed to compliment a user on his or her appearance, ask polite questions, and tell the occasional joke.
Noting that people respond favorably to compliments and negatively to criticism, Nass said that most messages you get from your computer today are the kind that say "You screwed up; You did terrible." Microsoft's stupid "Fatal Error!" and "Illegal Procedure!"--which I remember from my DOS days--spring immediately to mind, and messages that appear when Windows 95 or Word or Excel crash (which they do with notorious regularity) are little improvement.
(Sorry to inflict upon you my Microsoft-bashing mode again, but I've had the severe misfortune to be forced to work on a Windows 95 machine for the past two weeks. It has crashed--and I mean total lockup--at least once every day, and all I've been doing is running Word, Excel, and Project. I still plan on moving to Linux.)
But back to our story. As an experiment, Nass modified the spell-checker of a common word processor so that it would occasionally compliment the user if s/he spelt a difficult word correctly. The result was that "people thought our spell-checker was more accurate and smarter," even though the basic program was the same, said Nass. In a similar experiment using a program modified to tell a joke once in a while, users judged the program as being "smarter," or more intelligent.
And yet, "Almost every book on computer-human interaction says to avoid humor at all costs. But we like people more and think they're smarter if they're funny now and then." By ignoring the standard textbook advice and instead using such simple "tweaks,' users perceive the computer as being more intelligent, enjoys its company more, and is therefore more likely to get better use out of it.
Fundamentaly, the problem is that "We're dealing with 200,000 years
of human evolution. The human brain is not evolved for 20th-century media."
Indeed. Note, too, that we're dealing with notions of artificial intelligence
far from the usual complex stuff like neural nets. Some very minor--almost
cosmetic--changes to standard software can make it seem like AI. My view
is that if it feels like AI, smells like AI, and tastes like AI, then it's
AI!
Sir James Lighthill
Here are some excerpts from an obituary printed in The Times of London July 7:
Sir James Lighthill, mathematician and Provost of University College London, 1979-89, died while attempting to swim around Sark on July 17 aged 74. He was born on January 23, 1924. . . .
Considered by his peers to be one of the great mathematicians of the
century, perhaps even a genius, Lighthill was a pioneer in supersonic
aeronautics, in oceanographic studies and astrophysics. He virtually
created the field of biofluiddynamics, the study of how animals move through
air or water, as well as the study of the fluid mechanics of the cardiovascular
system. His ideas touched everything from earthquakes and the boundary
currents in the Indian Ocean to the movement of road traffic. He
held the senior mathematical chair at Cambridge, and became a leading adviser
on government scientific policy. . . .
Lighthill challenged the [British] Government to back research he had been engaged in with Post Office engineers and industrial scientists in the development of commercial television and communications satellites. Although Britain did not compete successfully in the space race, Lighthill was later the recipient, on his country's behalf, of two capsules of soil from the Moon, collected by unmanned Soviet spaceships.
In 1964 he became the Royal Society's resident professor at Imperial College, London, before returing to Trinity College, Cambridge, five years later as Lucasian Professor of Mathematics, a chair he held until 1979, when he was succeeded by Stephen Hawking. As well as continuing to publish on fluid dynamics - particularly the theory of waves in ocean and atmosphere - he worked on chaos theory and the unpredictability of large systems. . . .
In 1975 he produced a report for the Science Research Council, adjudicating on whether Britain should devote considerable resources to the development of artificial intelligence.
He was discouraging about the prospects for intelligent robots, and had grave doubts about any hopes of bridging the gap between man and machine. The work that had been done up to that time, he wrote, "casts doubts upon whether the whole concept of artificial intelligence as an integrated field of research is a valid one".
A great man, all the same.
Until
next week,
NEXT WEEK: Probably another issue of IMP.
Help Wanted: Got questions or comments on this article or on any other AI-related subject under the sun? Post it in the AIBB!