AgentSeeResearch Notebook
version 1.0.0 · created 2026-04-08 · updated 2026-04-08

D4: Two-Layer Machine (Understanding + State-Estimation)

design-argumentderivedoriginal
ClaimThe machine requires two layers (understanding + state-estimation), both necessary, neither sufficient alone: without state-estimation the understanding layer interacts with degraded humans without knowing it; without understanding the state-estimation layer cannot determine helpful vs harmful intervention.
This claim fails if
If a rule-based system with identical sensors matches outcomes on capacity maintenance and user-rated quality (P4, E3).

The machine requires two layers, both necessary, neither sufficient alone. The understanding layer (AI comprehension of values, context, narrative) enables the machine to recognize when current behavior diverges from long-term commitments and to contextualize observations. The state-estimation layer (real-time physiological + behavioral monitoring) enables the machine to detect when the human's capacity to access their own evaluative processes is degraded.

Without state-estimation, the understanding layer interacts with degraded humans without knowing it. Without understanding, the state-estimation layer can detect stress but cannot determine what constitutes helpful versus harmful intervention.

This design requirement is derived from the premises: if agency is state-dependent (F1) and state changes exceed unaided perception (F2), then an agency-supporting system must estimate state; if the objective is capacity for self-directed action (D2), then the system must understand what self-directed action means for this person.