Forms of thinking of autonomous intelligent agents: Features and problems of their organization

It is substantiated that, by analogy with living organisms, in order to solve in a problem environment of various complexity tasks of autonomous intelligent agents for various purposes, it is advisable to endow with three complementary forms of thinking: visual-effective, visual-figurative and conceptual thinking. The functional purpose of each type of thinking noted above has been determined. In particular, it is shown that visual-active thinking allows autonomous intellectual agents to function efficiently and purposefully and study the laws of transformation of the current situation of an a priori undescribed problem environment. Visual-figurative thinking tools are designed to output decisions in the process of planning behavior associated with a targeted transformation of the current situation of the problem environment based on a given knowledge representation model and decision inference procedures. Conceptual thinking serves to replenish the missing knowledge in the process of planning behavior in underdetermined conditions of functioning and to solve complex problems of behavior that require a change in situations of the problem environment to achieve a given goal. The main approaches, features and problems associated with the organization of tools for deriving solutions in the process of planning the behavior of autonomous intelligent agents based on the considered types of thinking are outlined.