The Strategic General
The Pentagon, always interested in the latest science, commissioned
a multi-million dollar research agenda to look at the potential uses
of artificial intelligence. The researchers were told
to investigate how artificial intelligence could be used for making
strategic decisions in a volatile environment.
In other words, a general-in-a-box.
Needless to say, this is not a simple problem. But after many years of work,
and many cost overruns, the research team had a something to demonstrate.
Being a new and snazzy technique, the Pentagon scheduled this to occur
in front of the Joint Chiefs of Staff. And to make it interesting,
the JCS would pose a problem to the "artificial general" on the spot.
The researchers set up their full suite of equipment (a massively
parallel processor, etc.) in the fanciest seminar room in the Pentagon.
With the JCS present, they speak in the following scenario
(voice recognition being one of the capabilities of their system):
Along with strident rhetoric out of Teheran, Iranian troops are massing
on the border with Iraq. The Saudis are nervous about Shiite rumblings
in their best oil producing provinces, and the Israelis are on high alert.
What should the U.S. response be?
After a few minutes (even for a massively parallel machine these problems can
be tough), the computer's speakers announce the following:
YES.
The lead researcher quickly explains that this system has real-time
conversational ability, so it is no matter at all to ask a follow up
question. He grabs the computer's microphone and asks:
Yes what?
After a few more minutes of churning, the computer announces the new answer:
YES, SIR.
|