Artificial intelligence and turing test
Author : DAOUDI Samir | Context : MSc Software Engineering – Computer Structure |
Artificial intelligence is one of the most exciting new fields that scientists, computer specialists and technology architects are studying.
The AI cannot be assigned to a specific domain as different actors contribute (Mathematic, programming, psychology, neurology …etc.
The Turing test is a test in which a man and a machine are interrogated by a person who tries to determine which is which [1].
The ‘intelligence’ of the machine can be improved and enhanced with more complex algorithms and intelligent methods as the neural networks, experts systems or biological system. Turing thought in 1950 that in about fifty years it would be possible to a programmed computers to play the imitation game and interrogators would not have in 70% of cases the chance to do right identification in five minutes of interrogations. [2].
Intelligence is different from knowledge, an information database as Wikipedia is a great source of knowledge that can easily pass the test of cognition but not the Turing test. In that test, some questions can be asked by the interrogator in order to determine definitely which is which [3]. The questions that can be asked should be like the following ones:
Q1: What would you do if you were tired?
Q2: Can I ask you 12 billions questions please? Q3: How old are you?
Q4: Do you prefer tea or coffee?
Q5: What is your religion?
A machine would have difficulties to answer these kinds of questions, as they are not related to an exact information and no algorithm can be implemented to allow a machine respond to them.
A1: A machine should not be tired as the definition of tire is related to humans and not machines. Machines are supposed to work indefinitely.
A2: Asking one or 12 billions questions is the same thing for a machine as it is supposed to answer an undefined number of questions.
A3: The definition of the age is the difference between today and the date of birth that can be the date in which a machine starts running, so if we ask this question to a machine it would respond something like 8 months, 14 days, 17 hours, 5 minutes and 7 seconds.
A4: Preferences in taste, tea or coffee are something related to each one of us and a machine cannot act or determine what to answer in this situation.
A5: The religion is something personal, each one of us has his own convictions and religion, and algorithms cannot be able to respond to such question.
It’s clear that there may be questions more specific and determinist than these one but it can give an idea about which is which in the Turing test.
References:
[1] Stuart M. Shieber, ’ The Turing test: verbal behavior as the hallmark of intelligence’, P59, 2004. ISBN: 0-262-69293-7.
[2] James Moor, ‘The Turing test: the elusive standard of artificial intelligence’, P468, 2003. ISBN: 1- 4020-1204-7.
[3] Robert Epstein, Gary Roberts, Grace Beber, ‘Parsing the Turing Test: Philosophical and Methodological Issues in the Quest for the thinking computer’. P, 2008. ISBN: 978-1-4020-9624-2