Human-in-the-Loop

Human-in-the-loop refers to an approach in which a person remains involved in an AI-driven process and can review, approve, modify, or override the system's output or actions.

The degree of human involvement can vary. A human may approve every decision, intervene only in exceptional cases, or monitor the system while it operates autonomously.

Human oversight is particularly important when AI systems affect individuals, make high-impact decisions, or perform actions that could create significant consequences.

Simply having a person somewhere in the process does not necessarily provide meaningful oversight. The person must have sufficient information, authority, and ability to intervene when necessary.