CONCEPT
Complex Adaptive System
Murray Gell-Mann’s foundational insight that the immune system, biological evolution, a child learning language, and a machine learning from data all share the same information-processing architecture—the scaffold that finally makes AI legible as a natural phenomenon rather than a categorical novelty.
A complex adaptive system, in Murray Gell-Mann’s formulation, is any system that does five things: acquires information from its environment, identifies regularities, compresses those regularities into a
schema, uses the schema to predict or prescribe behavior, and then revises the schema based on feedback from the world. The architecture is not a metaphor; Gell-Mann insisted with the precision of a physicist that it is the
same logic operating across radically different substrates—from the immune system to
large language models. This insistence dissolves the popular binary of “real intelligence” versus imitation, revealing instead differences of degree, substrate, and feedback mechanism. The framework’s power lies in the question it forces: not whether AI is “truly” intelligent, but
what kind of CAS it is—where its
schema compressions are reliable and where they are brittle, what shadows the compression casts, and what kind of human meta-schema is required to work productively with a system whose feedback mechanism