The article is devoted to the problem of constructing a knowledge system that, in the course of its development, will be able to approach a reliable representation of reality. Three possible ways of constructing a knowledge system are identified: the metaphysical way; the way of hierarchical models; and the way of isomorphism. The methodological basis of the study is general systems theory, including the author’s empirical-metaphysical general systems theory, which substantiates the possibility (and limitations) of the metaphysical path and the path of isomorphism for building a knowledge system, model theory, cognitive systems theory, etc. The main results of the research presented in the article are: identification of the genesis of the problem of providing arti-ficial intelligence systems that replenish the knowledge system with source data; identification of the limitations of the knowledge system formed when moving along the path of hierarchical models; analyzing the prospects of moving along the metaphysical path for the accumulation and generalization of deterministic knowledge formalized in the form of patterns; stating the high level of qualitative correspondence and adaptability of the knowledge system that can be constructed when moving along the path of isomorphism. The existence of ob-jective limitations necessitates the use of all three ways to achieve credibility, which is not possible in any one way.
Andrei Armovich Gribkov (Wed,) studied this question.