Human-in-the-loop
Human-in-the-loop actively involves people in an automated process. Useful review requires understandable information and ways to intervene.
Include a human decision in the process
Human-in-the-loop, or HITL, means a person actively participates in an automated system or process. They might correct a result or approve a proposed action. Their specific task must be defined.
In a fictional invoice process, the reviewer sees the original document, captured amount and matching order side by side. They can correct the capture or hold the case for clarification. A single confirmation button without that information does not provide useful control.
HITL can also occur in model training. However, correcting an item in a running process does not automatically retrain the model; that requires a separately established learning process. Human participation is also no guarantee that every error will be found.
Sources
Related terms
- Confidence scoreA confidence score assesses a model prediction according to a particular method. It does not grant business approval.
- Model cardA model card documents an AI model’s purpose, foundations, evaluation conditions and known limitations for a particular version.
- AI readinessAI readiness concerns the foundations of an AI project: objectives, usable data, skills, responsibilities and an operating workflow.
- Context engineeringContext engineering prepares information for individual model calls: relevant instructions, messages, sources and current states.
- Anomaly detectionMethods that flag unusual observations relative to a defined reference or a learned pattern.
- Forecasting modelA model that uses available information to estimate a future quantity for a defined period.
Articles on this topic
AI & Automation with ASES
AI assistants and automated workflows