|
|
|
|
A.I. and Unconscious From my personal point of view, A.I. research should avoid the error well represented by the famous Von Neumann's statement about our capability of making a computer do whatever we want by writing an appropriate program. There exist nowadays many examples of activities that a computer cannot do efficiently or absolutely, and that are normally carried out by human beings. In any case the fact that we can invent a software capable of implementing a given task does not show anything except that we are very good in writing programs. Apart these observations, one can state that the capability of designing an algorithm that performs a given task, does not necessarily implies that other information crunching systems implement the same algorithm to obtain the same results. Therefore the identification of a possible computer algorithm for a given activity does not necessarily clarifies how the same results can be obtained by the human brain. If someone exists convinced that, when we add 5 and 3 to obtain 8, what happens in our brain is well represented by the program stored in a pocket calculator, he/she has better to take a course of logic. Most, or all, the descriptions of human mind activities now available are given in terms of external meta-theories that capture, when possible, the equivalent algorithmic description of the mind specific performances. Confusing such meta-theories with the operating processes of information analysis actually going in the brain on can represent a very dangerous error in view of the understanding of the mammalian brain and mind operations. As a matter of fact, it seems reasonable to assume that the processes of information crunching that make the mind possible, take place at a non-linguistic level, that does not correspond to our meta-therotical models of possible reasoning. It turns then quite natural to identify such processes with what is normally identified by the term "unconscious". The differences between psychoanalytic unconscious and the implicit definition given above is evident and does not need a discussion. It would be interesting nevertheless investigating the possible relationships between these two concepts. A very good review paper by Max Velmans |