Abstract: Exploring the generation of intelligence in terms of Cognition Philosophy is a major challenge. This requires a conceptual framework that accounts for cognitive mechanisms, which is “adaptive representation”. Adaptive representation, as intrinsic mechanisms and explanatory categories of self-organizing systems at different levels, forms the basis of artificial intelligence moving towards generality, explainability, and reliability. This theory of adaptive representation for the generation of intelligence includes assumptions, inferences and principles as well as interactive processes of different hierarchical structures, aiming to show that the generation of intelligence is the result of the interactive emergence of different hierarchical structures of a self-organized entity or system through adaptive representation. Under the adaptive representation perspective, physical systems exhibit self-reaction and self-presentation of properties, biological systems exhibit self-adaptation and self-propagation of life, cognitive systems exhibit self-learning and self-expression, and AI systems exhibit machine learning and self-replication, and these different modes of representation precisely illustrate that adaptive representation is universal to all self-organized systems, and that the universal of general intelligence is adaptive representation, which implies that different domains of artificial intelligence have adaptive representational properties or functions, and constructing an AI system is creating an adaptive representational system.