The Reference Class Problem

Problem

Positioning against the wrong reference class -- comparing against systems as if they involve the same kind of machine at different settings. They do not.

JITAI systems are adaptive controllers. NASA systems are adaptive controllers. Schoeller describes robots and generic HRI. The proposed system is an intelligent agent that understands the human it serves. That is not a different setting on the same machine. It is a categorically different machine.

Correct framing

Existing engineering efforts provide foundational pieces (sensing, control theory, physiological monitoring, adaptive algorithms), but they all operate with machines that lack the understanding layer. The understanding layer changes what is possible at every point in the architecture.

The novelty question is not "has anyone specified this control topology?" It is "has anyone specified the integration of AI understanding with real-time physiological state estimation, oriented toward human capacity for self-directed action?" The answer is no, because the AI capability to do the understanding part at the required level has existed for approximately two to three years.

Additional reference class

The active inference psychotherapy literature (Chamberlin 2023 and broader family) uses FEP vocabulary and describes the same capacity-restoration topology. It must be acknowledged and positioned against. See active-inference-psychotherapy.md for full positioning.