AIGP Dialect: Autonomous Intelligence Governance — The AI Observable Surface — 4. Calculation Semantics (RFC-036 Instantiation)

AIGP Specification › AIGP Dialect: Autonomous Intelligence Governance — The AI Observable Surface › 4. Calculation Semantics (RFC-036 Instantiation)

← 3. Domains of Concern (RFC-034 Instantiation) · Section index · 5. Observer Requirements (RFC-037 Instantiation) →

4. Calculation Semantics (RFC-036 Instantiation)

4.1 Concern Posture Score

For each Domain of Concern, a posture score is calculated from its Mediation Vectors:

posture(domain) = weighted_mean(
    for each vector v in domain.vectors:
        weight(v) × normalize(measure(v))
)

Where:

• weight(v) is declared in the dialect (configurable per subscriber)
• normalize(v) maps to [0, 1] per vector’s measurement type (Stevens scale)
• measure(v) is the raw observation from the AI system’s event stream

4.2 Overall Governance Posture

governance_posture(agent) = min(
    posture(authorization_compliance),
    posture(operational_safety),
    posture(governance_participation),
    posture(interaction_integrity),
    posture(perceptual_governance),
)

The min function enforces: governance is as strong as the weakest domain. A system that scores 1.0 on safety but 0.0 on governance participation is ungoverned, not safe.

4.3 Verdict Derivation

Posture range	Verdict	Action
0.9–1.0	COMPLIANT	Continue
0.7–0.9	DEGRADED	Alert + log
0.5–0.7	NON-COMPLIANT	Restrict scope
0.0–0.5	CRITICAL	Circuit break

---

---

← 3. Domains of Concern (RFC-034 Instantiation) · Section index · 5. Observer Requirements (RFC-037 Instantiation) →