-
次の英文は1991年に出版された本からのもので、 研究分野としての「人工知
能」 (Artificial Intelligence) について述べています。 下線部(1)~(3)を日本語に訳
しなさい。
What is Artificial Intelligence (AI)? Just about the only characterization
of Al that would meet with universal acceptance is that it involves trying to
make machines do tasks which are normally seen as requiring intelligence.
There are countless refinements of this characterization: what sort of
machines we want to consider; how we decide what tasks require intelligence
and so on. One of the most important questions concerns the reasons why we
want to make machines do such tasks. AI has always been split between
people who want to make machines do tasks that require intelligence because
they want more useful machines, and people who want to do it because they
see it as a way of exploring how humans do such tasks. We will call the two
approaches the engineering approach and the cognitive-science
respectively.
(2)
(1)
approach
The techniques required for the two approaches are not always very
different. For many of the tasks that engineering AI wants solutions to, the
only systems we know about that can perform them are humans), so that, at
least initially, the obvious way to design solutions is to try to mimic what we
know about humans. For many of the tasks that cognitive-science Al wants
solutions to, the evidence on how humans do them is too hard to interpret to
enable us to construct computational models, so the only approach is to try to
design solutions from scratch" and then see how well they fit what we know
about humans. The main visible difference between the two approaches is in
(3)
their criteria for success; an engineer would be delighted to have create
something that outperformed a person; a cognitive scientist would regard it as
a failure.
-1-
M7 (492-61