B1 · Introduction to Artificial Intelligence

The Turing Test

The Turing Test, proposed by Alan Turing in 1950 as the imitation game, asks whether a machine can converse so well that a human judge can't reliably tell it from a person. The Chinese Room, proposed by philosopher John Searle in 1980, argues that passing such a test wouldn't prove real understanding.

8 min Beginner Jump to the activity

What you need: Nothing. This page starts from the beginning, with no programming and no formulas.

Definition

The Turing Test, proposed by Alan Turing in 1950 as the imitation game, asks whether a machine can converse so well that a human judge can't reliably tell it from a person. The Chinese Room, proposed by philosopher John Searle in 1980, argues that passing such a test wouldn't prove real understanding.

In simple words

You chat by text with two others, one a person and one a machine. If you can't tell which is which, the machine passes. Searle's reply: imagine someone in a room following a rule book to answer Chinese questions, without knowing any Chinese. The answers look perfect, yet no understanding is happening in there.

Why it matters

Modern chatbots have made this debate practical. They can often fool people in short chats, so does that make them intelligent? The discussion teaches you to separate seeming intelligent from being intelligent, and that matters when deciding how far to trust an AI.

How it works

How the test works

  1. A judge chats by text with a hidden person and a hidden machine.
  2. After a set time, the judge says which is the machine.
  3. If judges are wrong about as often as right, the machine passes.

Criticisms

  • It tests imitation, including imitating human mistakes.
  • It depends on the judge and the length of the chat.
  • It ignores vision, action and everything except conversation.
  • A system can fool people using tricks, without much intelligence.

Try it yourself

Play out the Turing Test

A judge chats by typing, and can't see who is answering. Follow one question.

Human or machine reply?

A judge asked: "What did you have for breakfast?" Guess who wrote each reply, then read why clues can mislead.

  • "Breakfast is the first meal of the day, eaten in the morning."

  • "Poha, with far too much lemon. My brother squeezed it."

  • "As an AI, I don't eat, but I can suggest a healthy breakfast."

  • "Skipped it, overslept. Don't tell my mum."

Common mistakes

  • Thinking that Turing claimed passing proves consciousness. He proposed a practical test in place of the vague question "can machines think?"
  • Treating fooling someone once as a pass. Serious versions need many judges and longer chats.

Check yourself

Question 1

1. The Turing Test judges a machine on:

Question 2

2. What does the Chinese Room argue?

Summary

  • Turing Test: can a judge tell the machine from a person in conversation?
  • Chinese Room: behaving correctly may not mean understanding.
  • Passing the test is about behaviour, and doesn't settle the bigger question.

Finished reading?

Sources: Turing, Computing Machinery and Intelligence (1950) · ARC Prize: ARC-AGI