AI Delegation
The governed transfer of decision-making authority from humans to AI agents — requiring clear scope, constraints, trust levels, and accountability.
AI delegation is the process of granting AI agents the authority to make decisions and take actions on behalf of the organisation. Like human delegation, it requires:
- Clear scope: what decisions is the agent authorised to make? - Defined constraints: what is the agent explicitly not allowed to do? - Appropriate trust level: how much autonomy does the agent have? - Accountability chain: who is responsible if the agent causes harm? - Revocability: can delegation be withdrawn, and under what circumstances?
AI delegation differs from human delegation in important ways: - AI agents can act at machine speed, amplifying both good and bad decisions - AI agents don't have judgment, intuition, or ethical reasoning in the human sense - AI agents follow patterns, which may not adapt to novel situations - AI delegation can be structurally enforced (governance gates), while human delegation relies on behaviour
How Constellation handles this
Constellation governs AI delegation through the Charter (defining what's delegated), progressive trust levels (how much autonomy), and the governance gate (enforcing boundaries). Delegation is explicit, versioned, and enforceable.