This work introduces the concept of mass–meaning decoupling, a computational framework for artificial intelligence that separates the physical existence of an entity (mass) from its context-dependent functional interpretation (meaning). Current AI systems primarily rely on statistical associations and fixed representations, lacking the ability to dynamically assign meaning based on context and intention. This paper proposes that meaning should be treated as a dynamic, computable state rather than a static label. The framework introduces semantic clusters, representing structured spaces of possible meanings, and a mechanism for dynamic semantic assignment based on context filtering, intention alignment, and evaluative scoring. A minimal formal model and an implementation architecture are presented, demonstrating how meaning-aware processing can be integrated into existing AI systems. The approach enables: context-aware interpretation of entities dynamic reassignment of meaning cross-domain semantic transfer more adaptive and flexible reasoning The paper further connects these ideas to modern language models, showing how they implicitly approximate context-dependent meaning, while extending them with an explicit semantic layer. This work proposes that general intelligence fundamentally requires the ability to assign, update, and recombine meaning across contexts, providing a foundation for more interpretable, adaptive, and ethically aligned artificial intelligence systems.
Dimitrios Moutsopoulos (Fri,) studied this question.