For a long time, sub-cognitive questions have been considered irrelevant in the Turing test. The reason is computers that have no experience in the environment as humans cannot pass in a controlled typical Turing test (French, 2000). However, this proposition has been proven to be untrue. Consider the following sub-cognitive questions that an examiner can ask the subjects.
- How decent is the term Flugly presented as a beautiful Hollywood actress?
- How respectable is Flugly presented in W.C. Fields movie as an accountant?
- How upright is Flugly being presented in teddy bear play?
Through the application of appropriate algorithms and simulations, the computer provides similar answers as humans. Thus, it becomes difficult for the examiner to identify which answer is provided by the computer. Computer algorithms generate answers depending on the way some words are synonymous. Consequently, the name Flugly will be well simulated in the last two questions (Turing, 1950).
Nonetheless, in the first case, the answer will be different. The reason is that the name Flugly or any other close word is not found in Hollywood vocabulary. However, Flugly is found in W.C. Field movies and the children play with teddy bears. The answers provided by the computer is consistent with that of human and the assessor can hardly guess whether the answer is from the machine or human (Turing, 1950).
Machines are deemed intelligent when they perform a precise task that human executes during communication. The interaction between the assessor and the human or machine subjects is done through the use of teletype (French, 2000). The major role of an assessor is to deduct whether an answer is provided by the machine or human. This will be attained via studying the results. The machine is considered intelligent under circumstances that the examiner cannot tell whether the answer given is from the machine or human.
References
French, R.M. (2000). Peeking behind the screen: the unsuspected power of the standard Turing Test. Journal of Experimental and Theoretical Artificial Intelligence, 12(6), 331-340.