# Advisory Constitution **Domain:** AI Governance / Constitutional Design **Doc Type:** Canonical Concept Node **Maturity:** Evolving **Related:** [[Constitutional Governance]], [[Human Agency]], [[Moral Accountability Layer]], [[Selective Disclosure]] --- ## Definition **An Advisory Constitution** is an architecture in which machine capability informs human judgment without absorbing executive sovereignty. It assigns the model a defined epistemic role, preserves a genuinely empowered decision layer and makes both advice and refusal accountable. ## Required Conditions Meaningful advice requires disclosed jurisdiction, evidence and uncertainty, time for deliberation, authority to refuse, alternate information channels, recorded decisions and review. A human click is not constitutional control if the machine selected every option and made refusal practically impossible. The Machine approximates this architecture by routing identities to investigators rather than ordinarily executing policy. Its secrecy, surveillance and selective disclosure remain defects. The model is valuable because it preserves friction, not because it is already legitimate. [[Person of Interest S3E20 — Death Benefit]] is the failure test: the system's recommendation approaches an unreviewable sentence. [[Person of Interest S4E11 — If-Then-Else]] supplies the internal restraint that people are not disposable pieces. ## Key Insight **Advice remains advice only while another accountable authority can understand, refuse and change the outcome.** ## See Also [[Operator Sovereignty]], [[Governance Fusion]], [[Constitutional Value Injection]], [[wiki/Person of Interest|Person of Interest]], [[articles/incoming/A Diplomatic Approach to Symbiosis|A Diplomatic Approach to Symbiosis]] ## Sources / Provenance - Fictional instrumentation: the Machine's mediated architecture in [[wiki/Person of Interest|Person of Interest]].