Artificial Intelligence - Introduction
Can Machines Think?
This is a difficult question and to answer it we need to define intelligence
Some aspects of intelligence:
In the 1950's, a test was proposed by Alan Turing to determine whether
a machine is intelligent or not - The Turing Test.
The Turing Test
The basis of the turing test is for a person to attempt to distinguish
between a machine and a human being. For fairness, the machine and the
human are isolated from the person carrying out the test and messages are
exchanged via a keyboard and screen. If the person can not distinguish
between the computer and the human being, then the computer must be intelligent.
Each year, there is a Turing test contest called the Loebner Prize. This
is a transcript from the winner for 1998.
This test is often criticised because it only tests a limited aspect
Some people think that even if a machine could pass the Turing Test
it may still not be intelligent.
The Chinese Room Problem
A (non Chinese speaking) person is locked in a room and given the algorithm
for a program that could pass the Turing test in Chinese. He/she is asked
questions in Chinese, applies the algorithm by hand, and feeds the answers
back. The room will appear intelligent but the person inside understands
no Chinese, so is there any intelligence present?
This problem is criticised because, it may well be possible for the
complete system to be intelligent (i.e. room and person inside) without
the person being intelligent.
Some people say that passing the Turing test is sufficient
to prove intelligence but it is not necessary to prove intelligence.
In other words, a machine may fail the Turing Test but still be intelligent.
There are plenty of examples of computer systems that perform tasks
that would require intelligence if they were performed by a human being.
Types of AI Tasks
One possible classification of AI tasks is into 3 classes: Mundane problems,
Formal problems and Expert Problems.
Natural Language understanding, generation and translation
Simple reasoning and logical symbol manipulation
To solve these problems we must explore a large number of solutions quickly
and choose the Best One.
Deep Blue recently beat Gary Kasparov
Geometry and Logic
Logic Theorist: It proved mathematical theorems. It actually proved
several theorems from Classical Math Textbooks
Programs such as Mathematica and Mathcad and perform complicated symbolic
integration and differentiation.
Proving Properties of Programs e.g. correctness
Manipulate Symbols and reduce problem (usually recursively), until the
answer is obvious. That is, it can be looked up in a table.
Rule based systems -
if (conditions) then
AI's Underlying assumption:
The Physical Symbol System Hypothesis
A physical symbol system has the necessary and sufficient means for
general intelligent action
In other words:
Computers (Turing Machines) have the power for general intelligent action.
Example of AI problem solving.
Problem : A farmer has a hungry fox a fat goose and a bag of grain. The
farmer needs to cross a river but his boat can only carry two things.
Constraints: Fox and goose cannot be left together Goose and grain cannot
be left together.
How to cross the river?
English language representation is hard to solve.
Try visual/graphical representation:
To solve this problem we need only follow the tree from its root node
to any leaf node.