Artificial Intelligence 1950

Computing Machinery and Intelligence

Alan Turing

Replace the unanswerable “can machines think?” with a test anyone can run.

Choose your version

In depth · the introduction

Turing took a foggy question — can a machine think? — and turned it into a game anyone can play: chat with two hidden partners, one human and one machine, and try to tell which is which.

The big idea

Asking whether a machine can “think” traps you in arguments about what thinking even means. Turing sidestepped all of it with a test. A human judge holds typed conversations with two hidden partners — one person, one machine — and tries to spot the machine. If the judge can't reliably tell them apart, then for every practical purpose, Turing said, we should call the machine intelligent.

It's a clever move, because it judges a mind by what it does rather than what it's made of. And Turing looked decades ahead: he argued that one flexible machine could be programmed to act like any other, and guessed that the smartest route to a thinking machine might be to build something simple — like a child — and let it learn from experience.

How it came about

Turing wrote this in 1950, while the very first electronic computers were flickering to life in a handful of university labs — and the field of “artificial intelligence” did not yet have a name; it would be coined six years later. He had already, in 1936, proved that a single “universal” machine could in principle compute anything computable, and had spent the war years breaking German codes at Bletchley Park. This paper, in the philosophy journal Mind, is where he turned from what machines can calculate to whether they can think.

He laid out his game with playful confidence, then spent much of the paper imagining — and demolishing — every objection he expected, from theology to mathematics to the claim that a machine can only ever do what it is told. It reads less like a technical report than a manifesto for a field that did not yet exist.

Why it mattered

This short paper gave artificial intelligence its founding question and its first benchmark. The “Turing test” became the popular measure of machine intelligence, and his hunch that a machine should learn rather than be fully pre-programmed is exactly how today's AI is built. He set the agenda decades before the technology could begin to meet it.

A way to picture it

Imagine texting with two strangers at once — one is your friend, the other is a chatbot — and you have five minutes to work out which is which, using nothing but the words on the screen. If the chatbot can keep you guessing, then as far as the conversation is concerned, it might as well be a mind. That texting game is, almost exactly, the test Turing dreamed up in 1950.

Where it sits

Turing's universal-machine idea, from 1936, is the reason a single computer can run any app at all. Here he takes the next step — from calculating to thinking — and plants the seed of machine learning. The line runs straight from this essay to the chatbots and large language models of today, which finally hold the kind of open-ended conversation Turing imagined.

The original document

Original source text

§1 · The Imitation Game

A. M. Turing · Mind LIX (1950): 433–460 · §1 The Imitation Game

I propose to consider the question, “Can machines think?”

This should begin with definitions of the meaning of the terms “machine” and “think.” … But this is absurd. Instead of attempting such a definition I shall replace the question by another, which is closely related to it and is expressed in relatively unambiguous words.

The new form of the problem can be described in terms of a game which we call the “imitation game.” It is played with three people, a man (A), a woman (B), and an interrogator (C) who may be of either sex. The interrogator stays in a room apart from the other two. The object of the game for the interrogator is to determine which of the other two is the man and which is the woman.

We now ask the question, “What will happen when a machine takes the part of A in this game?” Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman? These questions replace our original, “Can machines think?”

§5 · Universal Machines

§5 · Universality of Digital Computers

The digital computers considered in the last section may be classified amongst the “discrete-state machines.” These are the machines which move by sudden jumps or clicks from one quite definite state to another.

This special property of digital computers, that they can mimic any discrete-state machine, is described by saying that they are universal machines. The existence of machines with this property has the important consequence that, considerations of speed apart, it is unnecessary to design various new machines to do various computing processes.

They can all be done with one digital computer, suitably programmed for each case. It will be seen that as a consequence of this all digital computers are in a sense equivalent.

§6 · Contrary Views

§6 · Contrary Views on the Main Question

The Mathematical Objection

There are a number of results of mathematical logic which can be used to show that there are limitations to the powers of discrete-state machines. … The short answer to this argument is that although it is established that there are limitations to the powers of any particular machine, it has only been stated, without any sort of proof, that no such limitations apply to the human intellect.

Lady Lovelace's Objection

The Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform.

A variant of Lady Lovelace's objection states that a machine can “never do anything really new.” But who can be certain that the “original work” a man has done was not simply the growth of a seed planted in him by teaching, or the effect of following well-known general principles?

§7 · Learning Machines

Instead of trying to produce a programme to simulate the adult mind, why not rather try to produce one which simulates the child's? If this were then subjected to an appropriate course of education one would obtain the adult brain.

Presumably the child-brain is something like a note-book as one buys it from the stationer's: rather little mechanism, and lots of blank sheets. … Our hope is that there is so little mechanism in the child-brain that something like it can be easily programmed.

We can only see a short distance ahead, but we can see plenty there that needs to be done.