Problem of Computational Objectivation of Rationality in Artificial Intellectual Agents

Abstract

The subject of the article is objectification of rationality in artificial intelligent agents (AIA). The author considers two complementary trends in its context. The first ‘bottom-up’ trend is associated with attempts to artify rational reasoning and action in AIA, the second ‘top-down’ one is associated with attempts to interpret human thinking and behavior in machine terms. The first one is limited by the lack of semantic content, personal coloring and full-fledged psychosomatic embodiment of computational processes in AIA. Hypertrophy of logical normative components in them leads to mechanistic rigidity. The second is generated by the expansion of digital technologies into real life, their active fusion, which gives rise to the ideology of computationalism, according to which people are computing agents. At the same time, the convergence of the ‘intellectualization’ of AIA and the ‘machinization’ of human is a false appearance, since there are fundamental ontological and epistemological limits. The main ontological limit: human rationality is based on socio-cultural and bio-adaptive layers of the existence of a human subject, and therefore cannot be principally reproduced in ontologically simple, techno-physical AIA. The main epistemological limit: human thinking cannot be formalized and computationally objectified, since a meaningful semantic core is primary in it, which bears the stamp of subjectivity, and is also strongly linked to layers of implicit personal knowledge and other components of consciousness. Among them a number of sub- and non-rational phenomena stand out (emotions, values, common sense, morality), which are maximally distanced from algorithmic averaging, but without interaction with them rationality remains essentially reduced. It is concluded that the computational representation and objectification of definite components of rationality in AIA is possible, but is not rational itself, since their artificial separation from the larger, human-sized part of rationality initiates a split in it.