Human-in-the-Loop Control

Fact-check status

Reviewed claim by claim on 2026-09-02 against the primary or official sources identified inline as S1 and S2. Recommendations are design guidance unless a cited standard makes them normative.

Key jargon

Term	Plain-language meaning
Human in the loop (HITL)	A workflow that intentionally requires human input at a defined point. S1
Interrupt	A durable pause that preserves state while awaiting a decision. S1
Escalation	Routing a case to a person or higher authority when automated handling is unsuitable. S1
Four-eyes control	Requiring a second person to approve a sensitive action. S1

Key concepts

Place human gates at consequential decisions, not at every harmless model call. S1
An approval must show the proposed action, evidence, scope, and likely effect—not just an abstract yes/no prompt. S1

Concept map

flowchart LR
    A["Agent proposes consequential action"] --> B["Persist interrupt and evidence"]
    B --> C["Human approves edits or rejects"]
    C --> D["Resume or close"]

Diagram semantics verified against S1 and S2 on 2026-09-02.

A human gate works only when the reviewer sees the proposed action, target, evidence, alternatives, uncertainty, cost, reversibility, and policy decision. S1

Gate types

Gate	Purpose S1
Intake	Confirm task, authority, and data scope S1
Plan	Approve decomposition and intended tools S1
Exception	Decide on conflict, uncertainty, or scope expansion S1
Effect	Authorize a specific external change S1
Release	Accept quality/safety evidence before deployment S1
Incident	Stop, contain, rollback, and preserve evidence S1

Approval fatigue

Many low-information prompts train people to click “approve.” Batch harmless actions, reserve gates for meaningful transitions, and make rejection safe. S1

Exercise

Design an approval card for sending an email: sender identity, recipient, subject, body preview, attachments, data classification, source of content, and undo/recall limitations.