You do not currently have permission to view the full text for this document. Please sign in to an account with an active subscription or contact support.
Imagine you are in a room, in the role, not of analyst, but of the Interrogator. Behind two closed doors, respectively, are a man and a machine. With absolutely no contact or cues to prompt you, you are to determine by a series of typed questions and typed replies, silently exchanged between you and your invisible interviewees, which is human and which is a machine (or rather, an advanced computer simulating human intelligence). You may ask whatever questions you like, as few or as many as you like. The interrogation can last minutes or hours. It is over at that moment when—by a careful comparison of the two sets of replies—you are confident you can triumphandy differentiate between human and artificial intelligence.
If you would be willing to participate in such a thought experiment, you would in effect be participating in the celebrated Turing Test. Devised by the equally celebrated Alan Turing, forefather of computability theory, a mathematical genius who was chiefly responsible for cracking the German code in World War II, the Turing Test was proposed as an operational criterion of what success would be for a computer model of human intelligence.
Since Turing and the Turing Test, the movement (initiated in large part by him) called artificial intelligence, or AI, has come a long way. Today, computers play chess on the International Master level, converse and reason in correct, complex grammatical English, create elegant, sometimes aesthetically pleasing mathematical proofs, compose music, and diagnose illness.
This is not the same thing as simulating the wide range of human intelligence; and to spot a phony machine on the Turing Test—in other words, to be merely able to competently differentiate between human and nonhuman intelligence—you might say should be duck soup for an analyst, someone presumably schooled in the