Retrieval & Agents
Human in the Loop
HITLKeeping a person in the decision path to review, approve, or correct an AI action, especially for high-stakes steps.
Human-in-the-loop designs deliberately pause an automated system to let a person check or approve before it proceeds, rather than letting the AI act unchecked. For agents that can send emails, move money, or change records, a human approval step on the risky actions is a core safety and trust mechanism. It is also used in training, where human feedback and corrections improve the model over time.
Reference: Wikipedia: Human in the Loop
Related terms
AI Agent
An AI system that can take actions and use tools to pursue a goal, not just answer a single question.
Agentic AI
A general term for AI that acts with autonomy over multiple steps, rather than responding one prompt at a time.
Guardrails
Controls placed around a model to keep its inputs and outputs safe, on-topic, and within policy.
Reinforcement Learning from Human Feedback
A training method that tunes a model to be helpful by rewarding the responses human reviewers prefer.