Skip to content
AIGP

RFC-020: Governed Autonomy, Symbolic Intent, and D-DNA Evidence — 4. Design Thesis

AIGP Specification › RFC-020: Governed Autonomy, Symbolic Intent, and D-DNA Evidence › 4. Design Thesis

← 3. Problem Statement · Section index · 5. Terminology →

4. Design Thesis

This RFC is based on ten claims:

  1. Policy is no longer the primary control object for autonomous AI.
  2. Governance Intent is the primary control object.
  3. Natural language is insufficient as a runtime governance surface.
  4. Autonomous behavior must be constrained by symbolic, typed, machine-verifiable rules.
  5. Machines require architectural inhibition, not psychological deterrence.
  6. Enforcement state is temporal, scoped, expiring, and context-dependent.
  7. Communication to one affected actor must become verifiable communication to all affected actors.
  8. Human harm includes physical, cognitive, psychological, dignity, autonomy, and agency harms.
  9. D-DNA is the evidence genome of governed autonomy.
  10. Autonomous legitimacy must be replayable across time.

The primary doctrine is:

Centralize authority. Distribute enforcement. Broadcast state. Expire aggressively. Preserve D-DNA evidence.

← 3. Problem Statement · Section index · 5. Terminology →