On the problem of imitation of apperception processes by artificial intelligence systems

The article is devoted to the perception and data processing by artificial intelligence systems and the differences in information processing by a living brain and a machine. The authors note that the conceptual and abstract form of comprehension is significantly different from the natural reflection of the human mind, which makes it fundamentally impossible to imitate the processes of identifying objects by a computer. Purpose: to create an identification architecture that allows to imitate the process of living apperception that occurs in the human brain. Methods: presentation of information by which an object is identified in the format of an identification algorithm that takes into account all the data about the object that exists in the collective consciousness. Results: based on the existing views on the processes of artistic reflection and equating it to natural cognitive reflection, the authors conclude that it is incorrect to use the concept of “threshold function” that exists in programming and mathematics in existing models of imitation of thought processes in artificial intelligence systems. On the basis of the experience of analyzing figurative structures in literary texts, the authors argue that the mentioned sequence in the space of collective representations cannot be realized without taking into account even the theoretically possible minimum amounts of information involved in the formation of a virtual image of a given body. The need to develop a new information presentation architecture in artificial intelligence systems is stated, which implies the safety of all types and volumes of information of an identifiable object and is not focused on its relevance in specific cases. Practical significance: a conceptually new approach to understanding the essence of fixing and identifying information contained in the image of an object will make it possible to identify it in multitude of sensitive and emotional parameters, eliminating the possibility of an error inherent in statistical attribution methods. In addition, the complex perception of data at all levels of reflection within the boundaries of a specific sequence of processing separate information clusters suggests the possibility of independent determining the typological belonging of an object by an artificial intelligence system.