Distributed intelligence of multi-agent systems. Part I: Basic features and simple forms

This multi-part survey is devoted to the empirical characteristics and manifestations of intelligence as the capability of an autonomous agent to perceive, process, and use information in order to achieve its goal. Part I of the survey briefly describes the most significant attributes of “proto-intelligent” and intelligent systems, together with the main features of distributed intelligence (DI), individual human intelligence (HI), and artificial intelligence (AI). An inseparable connection is emphasized between the DI of human organizational and social systems and individual human intelligence. The simplest forms of “swarm” intelligence are considered as examples, and the key factors determining the effectiveness of DI in such multi-agent systems are presented, including the structure of interactions between agents, their collective goal-setting, external information recording, convolution, and processing, and the standard “images” of external influences. Their combination pushes the performance of a multi-agent system far beyond the capabilities of its individual agents. In part II of the survey, different forms of collective intelligence in human social systems will be analyzed and all known types of intelligence will be generally classified.