Multi-AI Checks and Balances: Layer-Appropriate Operation as an Architectural Paradigm
Derived from the Self-as-an-End Psychoanalytic Four-Layer Framework

Writing Declaration: This paper was co-drafted with Claude (Anthropic). All intellectual decisions, framework design, and final editorial judgments were made by the author.

Abstract

Current multi-AI collaboration architectures have made substantial progress in workflow orchestration, functional division of labor, and operational risk governance. What they have not yet structurally addressed is a different question: given a particular object (user, task, context), at what subject-level should the system operate? "Format this code" and "write me a letter declining a job offer" currently pass through the same pipeline — receive input, plan a path, invoke tools, return a result. This paper calls the missing dimension layer-appropriate operation: the system's capacity to recognize, given a particular object, at which layer it should operate, and to switch layers as objects change.

This paper derives, from the Self-as-an-End (SAE) psychoanalytic four-layer framework (12DD Id / 13DD Ego / 14DD Superego / 15DD Cert), an architectural paradigm orthogonal to functional division: layered checks and balances. Four Agents are not four workers assigned to different tasks but four operational modes of the same system — me-without-self (12DD), self-without-purpose (13DD), self-with-purpose (14DD), and self-with-non-dubito (15DD). The first theorem grounding the derivation is object-activation: the object determines the layer, not the role assignment. Inter-layer structural tension is maintained through remainder channels; intra-layer functional division is permitted and orthogonal. The paper derives three system pathology states, five non-trivial falsifiable predictions about currently unlayered architectures, and a general engineering implementation schema.

Keywords: Self-as-an-End, SAE, multi-AI collaboration, checks and balances, 12DD–15DD, psychoanalysis, object-activation, inter-layer dynamics, agent architecture, alignment, layer-appropriate operation