The Pros And Cons Of The Turing Test

Defining intelligence is a very difficult proposition and one which Alan Turing attempted to avoid answering as regards machine intelligence in the Imitation Game which has become known as the Turing Test (Turing, 1950). He posed the question “Can machines think?” which is he developed to ask if machines are able to converse in a way that can persuade humans they too are human. A machine is declared to have passed the test if human judges are unable to tell the difference between a human and a computer through a typed conversation. He suggested that a machine that persuades 70 per cent of human judges after five minutes of conversation should be deemed to have passed the test. The Turing test has become the most widely accepted test of artificial intelligence and the most influential. There are also considerable arguments that the Turing test is not enough to confirm intelligence. Legg and Hutter (2007) cite Block (1981) and Searle (1980) as arguing that a machine may appear intelligent by using a very large set of

To pass the test a machine will need to have a sophisticated knowledge of human behaviour to impress judges by adding quirks such as wrong answers and typing errors to appear more human. The capability to fake being human, rather than exhibiting true intelligence goes against the spirit of the Turing test (Legg and Hutter, 2007).

The Turing test is also fundamentally unreliable as it depends on human judges to decide on classifying a test subject as a human or machine. This has lead in some cases to unintelligent machines passing the test and a case of a human failing the test, according to Shieber (1994) as cited by Legg and Hutter (2007). The Eliza effect arguably demonstrates the unreliability of humans as judges of machine intellgence.

Other test for machine intelligence have also been devised such as the Loebner