Spread the love

Artificial Intelligence or AI is one the newest fields in science and engineering. Therefore a student in chemistry, physics or biology may feel that all the good ideas have already been taken by Galileo, Einstein or other famous scholars. On the other hand, Artificial Intelligence is very open to new Einsteins and Galileos.

What is Artificial Intelligence

There are several definitions of AI that we should consider among hundreds of them.

First one is, “The study of computations that make it possible to perceive, reason and act.”(Winston, 1992) (Thinking Rationally)

Second one is, “The art of creating machines that perform functions that require intelligence when performed by people.” (Kurzweil, 1990) (Acting Humanly)

Third one is, “The automation of activities that we associate with human thinking, activities such as desicion-making, problem solving, learning.. .”(Hellman, 1978) (Thinking Humanly)

Fourth one is, “Computational Intelligence is the study of the design of intelligent agents.”(Acting Rationally)

We defined four definitions as well as with 4 different approaches. Now, let’s take a look at those approaches.

AI Definition Approaches

1) Acting Humanly: The Turing test approach

Turing test was designed to satisfactory operational definition of intelligence. After writing down some questions, a computer can pass the test if a human questioner cannot tell whether the written responses come from a computer or from a person.

The computer should pass the following capabilities in order to pass the Turing test;

  1. Natural Language Processing: This is necessary to enable it to communicate successfully in English.
  2. Knowledge Representation: This is necessary for computer to store what it knows or hears.
  3. Automated Reasoning: This is necessary to use the stored information to answer the questions.
  4. Machine Learning: This is necessary to adapt to the new circumstances.

Total Turing Test

Turing test is avoided direct physical interaction among the questioner and the computer. Because physical interaction is not necessary for intelligence. However Total Turing test includes video signals that questioner can test the computers perceptual abilities.

The computer should pass the following capabilities in order to pass the Total Turing Test;

  1. Computer Vision: It is necessary to perceive the objects.
  2. Robotics: It is necessary to manipulate objects or move objects around.

It is more important to study the underlying principles of intelligence than duplicating one that already exists. As an example to that would be Wright brothers. They stopped imagining birds at some point and they started using wind tunnels and learning about aerodynamics. So Artificial Intelligence researchers do not study to built androids or robots. The main idea behind the AI studies is studying the nature of intelligence.

2) Thinking Humanly: The cognitive modeling approach

In order for us to say that a given program thinks like a human, we must have some knowledge on determining the way of how humans think.

There are 3 ways to determine that;

  1. Through Introspection: Trying to catch our own thoughts.
  2. Through psychological experiments: Observing a person in action
  3. Through brain imagining: Observing the brain in action

Once we have the precise theory of mind, it becomes possible to express this theory as a computer program.

The interdisciplinary field of Cognitive Science brings together computer models from Artificial Intelligence and experimental techniques from Psychology to construct and precise testable theories of human mind.

3) Thinking Rationally: The “laws of thought” Approach

Aristotle was one of the first to codify right thinking that is irrefutable reasoning process. His syllogisms  provided patterns for argument structures that always yielded correct conclusions. For example “Socrates is a man; all men are mortal; therefore, Socrates is mortal.” This study initiated Logic.

In this approach we are concerned about modeling uncertainty and dealing with complexity.

4) Acting Rationally: The rational agent approach

Agent is something that acts. Computer agents on the other hand, expected to do more, like operating autonomously, adapting a change or creating/following goals. Rational Agent acts to achieve the best outcome, or if there is uncertainty, the best expected outcome.

  • Agents can be organic, robotic or a software
  • Intelligent agents can operate autonomously in complex environments.
  • AI is concerned with the production of intelligent agents.

All the skills that we listed for the Turing Test also allow an agent to act rationally.  This approach has two advantages over the other three.

  1. It is more general than the “laws of thought”(Thinking rationally) approach.
  2. It is more suitable to scientific development than the other approaches. Because other approaches are based on human behavior or human thought.

References: http://aima.cs.berkeley.edu/ http://cs.lmu.edu/~ray/notes/introai/