Decision Legitimacy Framework
The Decision Legitimacy Framework studies what makes a decision legitimate in AI-mediated environments.
A legitimate decision requires more than a useful output. It requires a valid authority source, an accountable responsibility chain, a known automation boundary, and review conditions appropriate to the consequence of the action.
Output, Recommendation, Action, Decision
An output is generated material.
A recommendation proposes a possible course of action.
An action changes the state of the world.
A decision determines that an action should be taken, delayed, changed, or refused under an identifiable authority.
Confusing these layers creates judgment risk. AI can generate outputs and recommendations. Automated systems can execute actions. Neither fact alone establishes a legitimate decision.
Core Dimensions
- authority
- responsibility
- reversibility
- explicit delegation
- human review
- automation boundary
Framework Essays
- What Makes a Decision Legitimate
- When a Decision Is Not a Decision
- Limits of Automated Judgment
- Decision Without Explicit Authority Is Not Legitimate
- Automatic Execution With Irreversible Effects Is Not Legitimate
- AI-Generated Advice Does Not Transfer Decision Responsibility
- Responsibility Cannot Be Shifted by Default System Rules