This paper proposes a five‑level framework for organizing message generation and comprehension within thought and activity, aimed at strong language‑based artificial superintelligence (ASI).
The central claim is that a language model's context alone cannot expose the fundamental bases of utterance formation. Those bases are pre‑established by acquired schemas—systemic representations of the world of activity—and the truth‑connections accepted within them. Neither the context nor the model weights constitute deliberately constructed systemic representations.
The five levels are: (1) systemic representations of the activity world; (2) a particular situation extracted from those representations; (3) a narrative description of the situation; (4) the full sense of the message; (5) the text itself. A situation’s schema‑model is not identical to a story about that situation.
A recipient reconstructs the full sense from an abbreviated text by means of jointly built and agreed‑upon representations and collaborative activity. The reverse process proceeds from the full sense, through an expanded description and the situation’s schema‑model, back to the reconstruction of systemic foundations.
The theoretical grounding draws on G. P. Shchedrovitskiy’s systems methodology, A. Korzybski’s structural differential, and I. A. Mel'chuk’s Meaning‑Text theory.
The contribution is a specific systemic‑activity organization of content that enables examination, adjustment of message foundations, and control over the consequences of their transformation. Future work will focus on large graphical models of activity built on reconstructible systemic foundations, distinct from text generation solely from context.
Review