Glossary
Human in the Loop (HITL)
Human in the loop (HITL) is an oversight model in which a person can intervene in every decision cycle of an AI system, so that no output takes effect until a human has reviewed it and chosen to approve, change or reject it.
Term: Human in the Loop (HITL)
Human in the loop (HITL) is an oversight model in which a person can intervene in every decision cycle of an AI system, so that no output takes effect until a human has reviewed it and chosen to approve, change or reject it.
Why it matters
HITL is the best-known human oversight model and the most loosely used. The EU High-Level Expert Group’s Ethics Guidelines for Trustworthy AI describe it as the “capability for human intervention in every decision cycle of the system”, and caution that this is in many cases neither possible nor desirable.
We define the model by timing: the human acts before each decision takes effect. It suits decisions where a single wrong outcome matters, and it costs speed and scale.
How it works in practice
A local authority uses AI to recommend benefit decisions. Under HITL, a caseworker reviews each recommendation before it reaches the applicant. This works only when the reviewer has the authority, information and time to disagree with the system. A reviewer who approves by default provides no oversight; see automation bias.
Related terms: Human on the loop, Human in command and Automated decision-making. FAQ: Which Oversight Model Fits Which AI Decision?
Want this in context?
See how this term fits into the broader programme of work.