Collective Intelligence: From the Perspective of Complex Systems

Collective intelligence (CI) is widely recognized as a pivotal research frontier for the next generation of Artificial Intelligence. This paper presents a comprehensive review on the current status of CI research. First, we discuss the fundamental scientific challenges in collective intelligence and elaborate on the main theoretical approaches aimed at understanding consensus emergent phenomena and collective decision-making, drawing inspirations from biological swarm and social dynamics of human society. Second, we highlight four major CI and rapidly evolving technologies and application areas including multi-agent reinforcement learning, unmanned swarm system, open-source CI software development and federated learning. Last, we critically examine the limitations of current CI research efforts and identify promising future directions.