Return to MODULE PAGE

Artificial Intelligence: Can a Machine Think? (Page 1)

David Leech Anderson: Author
Robert T. Arrigo: Programming
Andy Schmidgall: Programming

Intelligence, computers, and robots

The word 'intelligence' is used frequently and with many different meanings. In fact, we have defined cognitive science as the study of intelligent systems. Using the term this way, it must pick out a concept broad enough to cover virtually any type of cognition. But even if the term is fairly broad and even if we sometimes speak as if intelligence admits of degrees (i.e., one thing can be more or less intelligent than another thing), it is still the case that intelligence is an important threshold term. That is to say, not everything qualifies as "intelligent". Some things (like humans and German shepherds) reach the threshold that is required for intelligence and some things don't (like rocks and vacuum cleaners). But what is it, exactly, that distinguishes those things that are intelligent, from those things that aren't? Soon we will consider answers to that question. But first there are some theoretical issues to review.

Remember that the theory of the mind that we have focused most attention on is functionalism (for a complete discussion, see Introduction to Functionalism). The job of a theory of the mind is to tell us the fundamental nature of mental states. Mental states are things like your "belief that it is Tuesday", my "hope that the Red Sox win a world series," and John's "headache pain". But what are beliefs, hopes and pains -- really? According to functionalism, mental states are functional states. Computers are themselves simply machines that implement functions. And so, according to functionalism, mental states are like the software states of a computer. In theory, then, it is possible that a machine running the right kind of computer program could have mental states, it could literally have a mind. It could have beliefs, hopes and pains. And, if it could have a mind, then it could also be intelligent.

All of this assumes, of course, that functionalism is true. But is it? Is it really possible for a machine to be intelligent? As we consider this question, keep in mind that a negative answer could give us reason to doubt the truth of functionalism; a positive answer would be consistent with its truth. Let's explore the question.

Is it possible for a machine to be intelligent? Well, lets ask that question of the machine that you are presently using to view this webpage. Could it be intelligent? To be so it would have to have the right kind of software. So let's give it some artificial intelligence (AI) software. We are now going to turn your computer into an "artificial agent" named Larry.

Larry does only one thing. He plays a game called "Last one loses". Since you will need to play the game with Larry, you need to know the rules. The game begins with ten pencils

THE RULES:

Last one loses can be played with any number of pencils (or any other kind of object for that matter). We use 10 pencils. On each move you must take at least one pencil and you can take no more than three. (Thus you have three legal moves: You may take 1 or 2 or 3 pencils.) The object of the game is to force your opponent to take the last pencil. The game is just that simple.


Your job now is to play a few games of last one loses with Larry. First play a few games with Larry going first. Then play a game or two where you go first. After you are done, return to this spot and continue reading.

So, did you play a few games with Larry? He's pretty good, right. In fact, if you let Larry go first, then he beat you every time. When you went first, how did you do? If you made a mistake on the first move, then Larry beat you again. But if your first move was the correct one, you may have been surprised to find that Larry FORFEITED! You might have thought:

"He shouldn't have done that. He might have been able to beat me if he hadn't quit."

While it is true that Larry might indeed have beat you if he hadn't quit, there's a reason why he quit. He knows that if you start and if you know what you are doing, there is no way for him to win. You see, he assumes that you are as smart as he is, that you know as much about the game as he does. There is a winning formula and if you know what that formula is, you are guaranteed to win as long as you get to go first. (NOTE: We could also call the winning formula a "winning algorithm." For a discussion of what an algorithm is see What is a computer?)

So, what do you say about Larry?

QUESTION 1: Is Larry intelligent?

In a typical group, only a few people are prepared to say that Larry is intelligent. But this is often a vocal minority. Some of the things that are offered to support a "Yes" (Y) answer to this question:

  • Y1:Larry is intelligent because If a human plays that well, we say that person is intelligent. Refusing to say the same thing about Larry is just prejudice (against machines).
  • Y2: Larry is intelligent at the game last one loses, even if he isn't intelligent in other areas.
  • Y3: Larry may not be as intelligent as we are, but he is intelligent. (Intelligence comes in degrees.)

When all is said and done, when you have studied all of the issues surrounding "machine intelligence", you may agree that Larry is intelligent. But then again, you may not. A majority of students in the classes that we teach judge that Larry is NOT intelligent. Most students are not willing to say Larry is intelligent. (Which doesn't, of course, mean that they are right, only that they are numerous.) What reason might one give to support the position that Larry is not intelligent? What answers might you give to the question:

QUESTION 2: What are the best reasons for believing that Larry is not intelligent?

Even if you believe that Larry is intelligent, it is still important for you to be able to explain the reasons that other people have for disagreeing with you. Take a minute to write your comments . . .

YOUR COMMENTS: Before you turn to the next page, jot down as many answers as you can to Question 2. If you are using this as a class assignment, bring your answer to class.