The classical test for computer intelligence is the Turing Test. This test was introduced by the mathematician and computer pioneer Alan Turing in 1950. In his paper „Computing Machinery and Intelligence“, he suggested a method for testing the intelligence of machines.

In this test, a human is having a conversation with a computer program via a keyboard and a monitor without intervisibility. If the human cannot distinguish whether a human or a computer program is at the other end, then the program may be seen as intelligent.

To pass this test, the difficulty lies mainly in speech recognition. There are some very good algorithms for speech recognition, when it comes to recognize the words from the acoustic signal. There are also quite useful algorithms to assign these sequences a corresponding grammatical structure. But to deduce the logical representation of meaning from these structures, that is something that is still very difficult.

Another development was to extend the man-machine communication to natural language. Computers should be able to understand not only what a person says, but also respond intelligently and aural.

An early, and still highly valued, representative of this category was ELIZA. Since then the field has changed considerably. There are a number of automatic agents and the clear separation between responses of men and machines is becoming increasingly difficult.

Apple’s Siri and Google’s Voice Actions is the recent step in this process.
Computers are smart for sure but are they intelligent?