Amid the interactions of Fourth Industrial Revolution technologies — including artificial intelligence, big data, the Internet of Things, virtualization, biotechnology and brain engineering — a growing number of new intelligent entities are emerging alongside existing biological ones, such as soft and hybrid intelligent entities and intelligent robots. On top of this, the formation of the Cyber-Physical-Social System (CPSS) is ushering in a "complex" new intelligence era, in which heterogeneous and nonlinear interactions among various types of intelligent entities are becoming routine. In this new era, a nation's social, economic and defense systems must solve problems such as efficient cooperation among components, the emergence of proper functionality of the overall system through such cooperation, adaptation or coevolution in response to changes in the environment or other components, and effective responses to uncertain future situations. To sustain and develop these systems, there is no choice but to seek solutions in a complex systems approach.
The simplest application of complex systems theory in the new intelligence era is so-called "systems thinking" — considering and handling "complex" social affairs in a manner suited to their nature. Examples include handling matters by considering the overall situation rather than fixating on components or parts of a system ("strategic thinking"); creating a positive system effect of "1+1>2" through the appropriate combination of components ("synergy effect"); and, in pursuing large-scale projects, taking a comprehensive and integrated approach based on the interconnections and interdependencies among each sub-project rather than handling them in isolation. Specifically, when pursuing a national and defense grand transformation strategy utilizing AI, this means not fixating solely on AI technology issues but pursuing the simultaneous development and convergence of other Fourth Industrial Revolution technologies, along with top-down design for innovation across all national domains — including the philosophy of science, administration, law, organization, education and training — and comprehensive project execution encompassing all government ministries. Other examples include designing joint combat systems with interoperability and jointness of multiple multi-domain manned and unmanned systems in mind, and preparing diverse scenarios that account for future uncertainty when formulating foreign policy. For such systems thinking to be utilized routinely in the real world going forward, it must move beyond its current form of being described in rigid engineering terms and develop more richly by combining with traditional wisdom and humanistic thought.
The second way to apply complex systems in the new intelligence era is to design and utilize new phenomena based on the principles of complex systems. Examples include the control and command of autonomous unmanned system swarms and the development of future new intelligence systems or operational theories. Behavioral control of autonomous unmanned system swarms can be seen in cases where metaheuristic algorithms — imitating the collective behavior of flocks of birds or schools of fish — are used to safely implement swarm flight or swarm submerged navigation. Existing examples of applying complex systems in intelligent system design include the deep neural network (DNN) theory that gave rise to large language models (LLMs), and neuromorphic chips that imitate the spiking neural networks (SNNs) of the human brain to simultaneously store and process data. Operational theories based on complex systems theory include "mosaic warfare" and "systems operations." Among these, mosaic warfare is a type of operation that constructs an adaptive system-of-systems centered on AI and network-based unmanned systems, and dynamically assigns and executes missions through automated and intelligent analysis of changing battlefield situations and mobilizable resources. This operational concept imitates the complexity of biological organisms, including adaptability to the environment, resistance to external intrusion, and the self-repair and self-management of cells. Systems operations theory, which emerged targeting the information warfare era, aims to create operational capabilities through the organic linkage of various operational components (units) composed in multiple layers.
The third application of complex systems theory is to analyze the emergence mechanisms of complex social phenomena and to control and manage those phenomena. The analysis of complex systems has traditionally made extensive use of modeling and simulation techniques using agent-based modeling or complex network theory. Recently, agent-based models are bringing innovation to areas such as international armed conflict analysis and real-time war simulation as they integrate with large language models and multimodal foundation models. Multilayer network theory has established itself as a methodology and tool for "mesoscience" theory, which seeks to reveal the process by which the microscopic world emerges into the macroscopic world, and in the military domain it is aiding the generation of combat power in the new intelligence era. In addition, various AI models that have recently emerged are themselves being utilized as tools to solve complex social phenomena. For example, graph machine learning is being widely used to predict the structures of various complex social systems in network form and to propose countermeasures, while deep reinforcement learning is being integrated with knowledge graphs or event graphs to support complex decision-making by humans or unmanned systems in specific environments. Various foundation models such as Chat-GPT (language and speech), DALL-E (vision), EVO-2 (DNA) and RynnBrain (robotics) — despite issues of hallucination — are demonstrating considerable capability in discovering complex patterns in the physical, biological and social domains and in proposing solutions to problems in each area.
Meanwhile, major powers are intensifying efforts to analyze, control and manage complex systems phenomena in the social, economic and defense domains by converging and utilizing various Fourth Industrial Revolution technologies. Representative examples are the theories of "digital twins" and "digital engineering," and of "parallel systems." Despite their differences, these theories aim to build a computer-based virtual world corresponding to the real world by utilizing AI, IoT, big data and simulation technologies, and to control and manage real-world phenomena and problems through automated and intelligent methods — by using real-world data in real time to conduct preemptive analysis, experimentation and simulation on the virtual world. Major powers have already deployed next-generation command and control systems applying this technology in actual operations.
As such, complex systems theory is drawing attention as a major methodology for solving complex social problems in the new intelligence era. However, general theories such as "open giant complex systems" or "complex adaptive systems" are "meta-theories" — theories for the sake of theories — and do not directly provide solutions to the concrete and diverse complexity problems of reality. Ultimately, it is important to develop complex systems theories and methods that match the changing circumstances of the times and the actual problems in each domain. In this process, as suggested by concepts such as mosaic warfare and parallel systems theory, the various complex system characteristics and intelligence phenomena found in nature and biological organisms, along with various Fourth Industrial Revolution technologies such as big data, IoT and simulation, will be materials and tools as important as AI technology itself.
