Objectivity and personal involvement in artificial intelligence: An examination of the Turing Test
Abstract
The aim of this report is to study what an intelligent system entails.
The study focuses on what makes natural intelligent system. This will be
important in that it will form a guideline for the construction of artificial
intelligent systems.
This study examines in-depth the Turing Test, and establishes the
difficulties which can be encountered by the strong follows of this test. The
study also goes further to illustrate an alternative test for intelligence in an
entity.
Turing demonstrated the test by describing a parlour game. He started
by saying that a man and a woman are in two separate rooms and
communicate with an interrogator only by means of a teletype. The
interrogator need to identify the man and the woman and in order to do so,
he/she may ask any question capable of being transmitted by teletype. So, the
>:
man tries to convince the interrogator that he is the woman, while the woman
tries to communicate her real identity. At a given point in time during the
game, the man is replaced by a machine. if the interrogator remains unable to
distinguish the machine from the woman, the machine will be said to have
passed the test and we will say that the machine is intelligent.
This study identifies three limitations with the Turing Test:
First is the Achievement Problem. By the time Turing was proposing
this test, AI was at its infancy. Then, Turing went ahead to propose contest
between the infant AI (machine) and the human person who had at that time
really developed his/her intelligence to quite a high level.
Second Problem is The Fuzzism of Intelligence. It is not true that for
any subject to be termed as intelligence, it has to defeat a human person's
intelligence. Other animals exercise intelligence too, for instance a baboon
that peals a banana before eating it.
Third is The Problem of Missing Standards. According to Turing
Test, there are some essential features, which ought to have been included,
and which may lead the whole exercise to a complexion. For instance there
was the need of standardization of the test situation, which he failed to do.
These limitations are discussed in details later on and an alternative
solution test provided to overcome these limitations.
My approach for the test of the intelligence of an entity divides the test
into two levels. The first level looks at an entity and its constructional
components. The second level analyses the action of the entity that is usually
manifested by its components.