Governance

Human‑in‑the‑loop

Human oversight modality allowing human intervention in AI decisions.

Definitions (3)

A human oversight arrangement where a person is actively involved in the AI system’s decision‑making cycle (including review, correction, or approval), encompassing related modalities such as human‑on‑the‑loop and human‑in‑control to ensure appropriate oversight.

A governance practice where a teacher or school staff reviews, moderates, or validates AI-generated outputs prior to dissemination to learners, ensuring pedagogical alignment and preventing harms. The guide emphasizes this as a core mitigation for higher‑risk uses like assessment.

Operational requirement that a human reviews, validates, or authorises AI-generated outputs before they are acted upon in high‑risk or critical public‑service contexts, ensuring accountability and mitigating harm from erroneous or biased outputs.