The pioneers: Walter Pitts, Warren McCulloch y Alan Turing
It is no longer uncommon to talk about a machine's ability to reason like a human, or about robots that perform dangerous tasks. Science fiction had already opened our minds to the potential of Artificial Intelligence. As recently as 1942 Isaac Asimov wrote the book Runaround, where we can read the now famous Laws of Robotics. But science fiction is simply that, fiction. If we want to know who the true founders of Artificial Intelligence are, we need only look at the year after Asimov’s book. In 1943 Pitts and McCulloch suggested the construction of a computer based on a biological neural network, while Alan Turing developed the test, which now bears his name, for judging machine intelligence. If the human taking the test is unable to detect whether the terminal is human or machine, the system’s intelligence would therefore be proved.
Learning Machines Vs iterative development
It is not easy to define this multidisciplinary area which, through various sciences (mathematics, computer science, logic, etc.), aims to solve everyday problems autonomously by imitating human behaviour.
The notion of Artificial Intelligence is divided into two schools of thought: Conventional A.I., which is characterized by formalism and statistical analysis applied in its methods (expert systems, case-based reasoning, Bayesian networks); and computing A.I., which involves an iterative learning based on empirical data (neural networks, fuzzy systems and evolutionary computation), that can be applied in many fields.
One must look beyond the concept to understand that it not only brings together systems that think like humans; A.I. is as complex as the neural networks it aspires to resemble. Systems that think with the same logic as a human and improve reasoning to find an optimal solution, and systems capable of perceiving the environment, also form part of this field.