The Chinese Room

The Turing Test: The Imitation Game

A simpler question is:

Can a machine win at the imitation game?

The Imitation Game

Imagine the following setup:

1. You sit down at a computer.

2. The computer is connected to two other computers, A and B, located in a separate room.

3. Computer A is controlled by a person.

4. Computer B is controlled by a computer program, or machine.

5. You type questions to A and B, and they answer.

6. If you can tell which one is the person and which one is the program, you win.

7. If you cannot tell which one is the person and which one is the program, the machine wins.

If the machine wins the imitation game, then, according to this test, it can think like a human thinks.

The Chinese Room Argument

The Chinese Room

Imagine the following setup:

1. You sit down in a room with a book filled with instructions in English.

2. A piece of paper with lines and squiggles slides under the door.

3. You take the piece of paper and, following the instructions in the book, create a new set of lines and squiggles on a new piece of paper.

4. You slide that new piece of paper out under the door.

5. It turns out that the paper slid into the room is a question written in Chinese.

6. The paper you slide out of the room is the answer to that question written in Chinese.

The person asking the questions and reading the answers would have no idea that you were not a native Chinese speaker.

But if they knew the truth, there is no way they would believe that you actually knew Chinese.

The Argument

The Chinese Room is a version of the imitation game.

In the Chinese Room, the person inside the room produces the correct answers without understanding Chinese.

So we would not say that the person in the room knows Chinese just because they can follow a program that produces the correct answers.

But if that is true, why would we say that a computer can think like a human just because it produces the correct answers?

Questions

1. What is a machine?

   • Are brains machines?

2. Can we know how to do something without knowing that we are able to do it?

   • Can we know how to do something without being able to accurately explain how we do it?
   • How do you make a left turn on a bike? You know how to do it, but the answer you are likely to give is probably wrong.

3. Searle seems to say that we should be studying consciousness and not cognition. Is that true?

4. What difference does it make if machines can think?

5. Should the way we treat a machine change if it turns out that it is a thinking machine?