Human-in-the-Loop

A design paradigm where human judgment is integrated into the AI decision-making process, allowing humans to review, approve or override automated decisions.

A person reviews and approves the AI's decisions before they take effect. For example, an AI flags potentially fraudulent transactions, but a human analyst makes the final call.

Governance frameworks must define where human oversight is required in AI decision-making. For high-risk applications, human-in-the-loop processes are often a regulatory requirement and a critical risk mitigation strategy.