Computers Why the Turing test is a poor gauge for todays artificial intelligences / – my slice of internet / Oradea, Bihor, Romania.

You might have noticed the huge progess of the computer processing power and the development of the programming algorithms in the laste decades, especially if you have lived them. This evolution of technology made possible the appearance of the Artificial Intelligence, pieces of complex software that can interact with their environment, learn from its changes and their own actions. In one word: they adapt. Some examples are the supercomputers that are masters in chess, walking computers like Asimov from Honda (perhaps the most famous one), rolling robots that have a lof of tasks to acomplish in labs with only one target in mind: further development of their algorithms and their inteligence.

As the definition says, the Turing test is a test of machine’s ability to exhibit intelligent behaviour, a criterion proposed by Alan Turing for deciding whethever a computer is intelligent. Even if it can be considered that if a computer acts, reacts and interacts like a sentient being that it is sentient, the Turing test also states that to be considered sentient an Artificial Intelligence must acheive a 30% succes rate, that means it must convince a human that he is talking to another huma in 30% of all the trials. So, if a computer’s conversation is indistiguishable from a human’s conversation and it can fool a human for some time than the machine and its Artificial Intelligence can be credited with some intelligence.