Nothing is simpler than the structure and conduct of the Turing test: the test is based on the principle of questions and answers between a (human) interrogator and two anonymous respondents whom the latter does not see. The questions are free and are not determined in advance. The interrogator asks them to his interlocutors via an input tool, such as a keyboard or a screen, without benefiting from any visual or auditory contact with his interlocutors. If the human interrogator cannot, based on the answers, determine which of the two respondents is a machine, then it is possible to consider the intelligence of the latter as similar or equal to that of a human being.
As of today (i.e. March 2022), it is impossible to cite examples of machines that have officially passed the Turing test. Nevertheless, this experimental device still remains essential to the development of artificial intelligenceparticularly in the context of deep learning, reinforcement learning, supervised learning or unsupervised learning.
In the future, the mode of communication of machines based on artificial neural networks and comparable to ours will not only have a primordial role to play on social networks and in the context of customer service; the communication based on artificial intelligence will also be increasingly present in areas such as medicine, diagnostics, agriculture, security, surveillance, marketing, transport and production.
