Definition
The Turing test checks whether a computer can chat so naturally that people can't tell it apart from a human. In 1950, British mathematician Alan Turing proposed it as a way to replace the vague question "Can machines think?" with something you can actually test.
A Chat Behind the Curtain
Picture two chat partners hidden behind a screen. You don't know their faces or voices. In one room sits an ordinary person. In the other, an AI program is plugged in.
A judge talks freely with both, using nothing but text messages. The judge can ask anything: light small talk like today's weather or what's for lunch, or deeper questions about favorite music or life values.
After a good long chat, if the judge can't correctly say which one is the computer and which one is the human, the machine passes. Looks and voice are hidden, so the machine is judged on natural language skills alone.
Notice what this design does: it takes appearance out of the picture. You won't be wowed by a shiny steel robot or flashy hardware. The only yardstick is the words going back and forth.
To Be a Bit More Precise
To be more precise, the Turing test doesn't check whether a machine has a real mind, like a soul or a sense of self. It's a behaviorist test: it looks at how well a machine imitates human behavior and conversation.
Here's a funny twist: a computer that acts too smart can actually fail. If it solves a 10-digit multiplication in 0.001 seconds or recites an entire encyclopedia, the judge will instantly think, "That's no human."
So a machine taking the test may make typos on purpose, or wait a few seconds before replying, acting out human slip-ups. Seeming tired or tossing out a silly joke counts as part of smart imitation.
In the end, instead of the hard question "Does the machine truly think?", this practical test focuses on whether it can interact at a level you can't tell apart from a person.
Limits and Meaning in the Age of Modern AI
Today's widely used large language models (LLMs) write sentences as fluently as people do. In a 2025 five-minute chat experiment by U.S. researchers, one language model was picked as the human more often than the real humans were. But people disagree on whether such a result means machines have reached human-level intelligence.
There's also debate over whether giving a convincing answer is the same as understanding the real-world principles behind the words. That's why, these days, tools that check logical reasoning and complex problem-solving ability are used alongside conversation tests.
The question of whether imitating conversation amounts to real understanding has been a criticism ever since the test appeared. The best-known example is philosopher John Searle's 1980 "Chinese Room" argument.
Still, the Turing test holds a unique place in AI history. It pulled the philosophical puzzle "Can machines think?" down into the realm of a concrete experiment that anyone can see and verify.
Common misconceptions
A machine that passes the Turing test has a self and thinks for itself like a human.
The Turing test only rates how well a machine imitates conversational behavior. It can't prove that real thoughts or feelings exist inside the machine.
Where you meet it
You're messaging with a customer service chatbot and can hardly tell whether it's a real person or a program.
ELIZA, an early chat program built in 1966, echoed back what people said and kept conversations going like a professional counselor.
In one sentence
The Turing test uses text chat from behind a screen to check whether a machine can imitate human conversation so well that you can't tell it from a person.