The Turing Test

What is now called the Turing Test was proposed by mathematician and early computer scientist Alan Mathis Turing in 1950¹. The central question this test attempted to solve was how we could know, or when we should say that a machine was capable of humanlike thinking.  At the time Turing believed that in 50 years computer would pass his test with regularity.  That was 60 years ago and we are, arguably, still not to this point.

The Turing Test is based on “the imitation game”.  The imitation game is played by three people, two subjects (one male, one female) and one interrogator (groups of participates may serve as the interrogator).  The subjects are separated from each other and the interrogator.  It is the job of the interrogator to ask questions of the subjects and determine who is male and who is female based on their answers.  There’s a catch however.  One subject is trying to help the interrogator guess correctly, while the other is trying to make the integrator err.  This is not to say of the subjects that one must tell the truth and the other must lie (if this were the case there is a single, simple question to solve the problem every time).  Success as an interrogator depend on one’s ability to understand and play on sex stereotypes, but clever subjects can also make use of these stereotypes to foil the interrogator.  A clever interrogator faced with equal clever subjects will, in theory, only guess correctly about 50% of the time.

The Turing test keeps the role of the interrogator but replaces both subjects with a single computer.  The interrogator must decide if the subject is a fellow human being or a computer.  If, Turing claims, the computer can fool the human interrogator about 50% of the time we have reason to believe the computer is a thinking think as is a human being.

Why should we think this?  Computers are capable of many difficult and impressive tasks well beyond the capabilities of a bare human mind.  Adding two 7-digit numbers is easy enough for a human with a pen, paper and knowledge of basic arithmetic, but a computer, even a primitive one, can solve this is a vanishingly tiny fraction of time.  On the other hand, some tasks that humans find comparatively simple are beyond a computer’s abilities.  Suppose I give you the following task:

Step 1- add the numbers 1 and 2.

Step 2- add 2 to the result.

Step 3- repeat step 2 until a result that is evenly divisible by 2 is reached (i.e. – the result is an even number)

You will not take you long to realize that there is not such result and the answer after steps 2 and 3 will always be and odd number.  A computer on the other hand will not realize this and will continue adding 2 and checking the result until either its working memory reaches capacity or a pre-programed number of attempts is made and the process is halted without yielding an answer.

Additionally, how might you reply to the following inquiry:

“Did you hear about the Irishman who found a magic lamp? When he rubbed it a genie appeared and granted him three wishes. “I’ll have a pint of Guiness!” the Irishman replied and immediately it appeared. The Irishman eagerly set to sipping and then gulping, but the level of Guiness in the glass was always magically restored. After a while the genie became impatient. “Well, what about your second wish?” he asked. Replied the Irishman between gulps, “Oh well, I guess I’ll have another one of these. … I told you the joke because I want you to explain it to me.”²

You might be able to say something about stereotypes of the Irish or the illogic of such a request, but the more general question: “why is this funny” is something even humans have a hard time answering.  Would a computer do any better? 

Computers are machines that manipulate symbols and use brute force to answer solve complex problems. Deep Blue, one of the first chess playing computer capable of beating grandmaster human players, can do so by looking farther and farther ahead in the game then any human player is capable.  Humans can consider, at best, three or four moves per second, while Deep Blue can consider billions of possibilities in that same slice of time.  Watson, the super-computer that beat both Ken Jennings and Brad Rutter³ at the game show Jeopardy!, did so by having a super-fast, stable internet connection and programming for interpreting results of its queries.  This is a far cry from the wetware resources and institutions of the human competitors.

Number crunching and symbol manipulation certainly part of human thinking but this is not the whole story.  Human thinking requires human understanding.  To make someone believe you are someone other than who you are requires a cognitive tool kit beyond what computers have to offer.  So if a computer can accomplish a task that previously only human mental powers could achieve, why would we not say it was engaged in the act of thinking?