Information Technology Reference
In-Depth Information
fied, so intelligence clearly is not directly linked to the ability to use
a workstation. Overall, various intelligent life forms must allow a
more diverse range of interfaces than are typically included in state
ments of the Turing Test, and this points out some shortcomings.
A second type of criticism suggests that the Turing Test is too in
clusive, because it does not consider how the computer or other respon
dent formulates the answers. Perhaps the most wellknown of these
critics is philosopher John Searle, who puts forth an objection called
the Chinese Room. In this scenario, a person is placed in a room and
asked to translate from Chinese to English. This person knows nothing
about Chinese, but is given a collection of rules to follow. When this
person receives something to translate, the person looks at the charac
ters on the page, identifies the relevant rules, applies the directions
given, and produces a translation. Using this method, the person would
pass the Turing Test (at least with regard to the translation of Chinese
to English) without actually demonstrating any intelligence. Effectively,
this objection states that blackbox testing cannot provide an adequate
mechanism to determine if an entity has intelligence; a similar argu
ment might well apply to any blackbox intelligence test.
One counterargument to the Chinese Room objection analyzes
functioning in the human brain—at least as we understand the
thought process. Following the scenario for translation from Chinese
to English, a person performing translation receives input as Chinese
characters and speaks the output in English. To actually receive this
input, the human translator reads the characters with her or his eyes,
so this information becomes a sequence of impulses that stimulate
neurons—first in the translator's eyes, then in the neurons of the brain,
and finally in signals to the mouth, tongue, vocal chords, and so on to
produce speech. In this process, each neuron gathers electrochemical
signals from its neighbors. Then, if conditions are right, a neuron fires,
adding further electrochemical signals to the human's system. In this
respect, the neuron in the translator's brain has much the same role as
the human in Searle's Chinese Room. Neither knows what they are
doing; they simply follow rules in their environments. Neither has real
knowledge of Chinese or English. From this standpoint, the process
for a human translator depends upon basic neurons that respond to
electrochemical stimuli, just as the process for the person in the
Chinese Room follows another collection of rules. From this perspec
tive, it is difficult to know just what processes we should be examining
in identifying intelligence, and blackbox testing may seem appropri
ate after all.
