What is a Turing Test?
A Turing test is about whether a computer has equivalent reasoning ability to a human. Originally called the "imitation game" and developed by Alan Turing in 1950, the test became more precise and concrete after AI became an academic discipline in its own right. Artificial intelligence allows the machine to think for itself. A human questioner is asked to start a conversation with two completely unknown interlocutors via a screen and a keyboard, quite simply without visual or auditory contact. One interlocutor is a human being and the other is a machine. If, after intensive questioning, the questioner cannot tell which of the two is the machine, then this machine has absolutely passed the Turing test. The machine can be said to have a thinking capacity comparable to that of a human being.
What are the criticisms of the Turing Test?
There is criticism of the Turing test regarding the duration of a test. A computer can be really good when it comes to answers that can only be given as yes or no. Such answers only relate to a specific range of knowledge, such as mathematical number theory. In contrast, when it comes to a topic that is part of a broad general conversation, the computer's answer is not comparable to real people. Especially with emotional topics or social issues, you can't yet compare the computer to real people. On the other hand, it is the case that a Google search engine answers in much more detail than any normal human person.
Which systems have been tested so far?
Among many other systems, there is the chatbot Eugene Goostman, which was the first system to pass the Turing test in 2014. In the summer of 2017, an AI capable of independently writing reviews was presented at the University of Chicago. The machine-generated reviews were quite simply presented with human-generated reviews to some 600 test subjects for evaluation. In the blind test, the computer's reviews were judged to be similarly useful and thus it passed the Turing test.
As early as 1956, much was said about Artificial intelligence thought. Programmes like LOGIC THEORIST with which 38 theorems from Russel and Whitehead's fundamental work Principia Mathematica could be proven. It could thus be shown that computers can effectively process not only numbers but also symbols.