As you know, science does not fully know how the human brain works. A similar situation is observed with AI: the mechanism according to which the neural network makes decisions is not completely clear. “A neural network is not a program, it is a specially organized data processing process, during which certain patterns are formed inside an artificial neural network - the same as in a person’s brain. But how they are formed, a person cannot understand, because they do not appear in the usual language of formulas or concepts, but they appear on their own foundations of a neural network that correlates factors with each other. And for a person, this is all a “black box”, he does not know how it is done, everything happens, in fact, blindly,” notes Mr. Kuznetsov. But, it would seem, how can AI be delegated to make important decisions if it is not known how the neural network makes them? “According to my observations, few people are interested in this now.
Since artificial neural networks demonstrate their effectiveness, it means that they should be implemented wherever possible. And, as you know, there is no time for buildup. And the fact that an artificial neural network is a cybernetic "black box" does not matter. There will be problems - we will solve them. It seems to me that many people think so, ”says Yuri Cherny, head of the Center for the Study of Informatics Problems of the INION RAS, in an interview with Kultura. Academician of the American Academy of Sciences Alexander Kuleshov, rector of the Skolkovo Institute of Science and Technology, believes that by actively using artificial neural networks, we returned to the 16th century, “when scientists observed some kind of effect, but could not explain it.” That is, in fact, there was a return to the pre-Galilean era - to the times of natural philosophy.
But it is always useful to look at the past, because then European science was being formed. The philosopher and mathematician Rene Descartes is considered to be the ancestor of the approach from which modern scientific methodology has grown, and, in fact, artificial intelligence. He believed that any fact should be questioned, which, of course, meant the constant presence of consciousness. This worldview put the human intellect at the center of the world, which could cognize this world. “There are several basic concepts that are shared by many scientists today - that a person has no consciousness, that a person is an animal, and the Universe is a big computer. There is also a more radical, but gaining popularity, point of view that the Universe does not exist, just as reality does not exist, ”writer, futurist, artificial intelligence expert Igor Shnurenko tells Culture. For Descartes, human consciousness lay at the basis of the knowledge of the world, the Universe, but now a person is gradually disappearing from this scheme.
Such a worldview is called an object-oriented ontology, which puts not a person in the center of perception, but any object. At the origins of this concept is, according to Mr. Shnurenko, postmodern philosophy. Mr. Pivovarov formulates the main provisions of the scientific community as follows: “The first thesis is that we recognize the existence of consciousness, the second is that consciousness is inextricably linked with the brain, and the third is that apparently there are mechanisms that we do not understand, related to the functioning of neurons, neural ensembles, how they work. The scientific community believes that consciousness is caused by the work of a large neural network in the brain, arranged in a special way. This means that theoretically we can reproduce this consciousness in a computer if we adequately transfer the work of this network to it.
Some scientists and philosophers agree with this. And the other part - no, believing that there is no evidence that consciousness is caused by the interaction of only neurons, so it is not at all certain that we will recreate it. .