Mehmet EmrahKonyaTR
← Back to glossary

Human-in-the-loop

Döngüde insan

D6

In the cited HITL NLP survey, human-in-the-loop is an arrangement that integrates human feedback into a model-development or deployment workflow.

Review status: 2027-02-24

Technical explanation

The relevant human interaction can vary by task, goal, feedback method, and the point at which feedback enters the loop.

Conceptual boundaries

Human-in-the-loop is not synonymous with all human oversight or human–AI collaboration, and it does not guarantee accuracy or AI safety.

Provider-neutral example

A content-routing system can require a reviewer to approve or reject a proposed category before the item is released to a public queue.

Limitations

The cited HITL research varies in tasks, goals, human interactions, and feedback-learning methods, so it does not establish one effective arrangement for every system.

Related concepts

Atomic claims and evidence

  1. 1.1NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0) — D6 human and organizational slice
    Source
    NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0) — D6 human and organizational slice
    Source role
    Authoritative source
    Exact locator
    Appendix C, printed p. 41
    Supported claim
    NIST discusses Human-AI team configurations together with the roles and responsibilities of humans overseeing AI-system performance.
    Last verification
    Review due
    Scope limitation
    This does not define one mandatory loop position, approval mechanism, or operational label for every AI system.
  2. 2.1Regulation (EU) 2024/1689 (Artificial Intelligence Act), consolidated text of 27 July 2026 — D6 human-role slice
    Source
    Regulation (EU) 2024/1689 (Artificial Intelligence Act), consolidated text of 27 July 2026 — D6 human-role slice
    Source role
    Authoritative source
    Exact locator
    Article 14(1)-(4), current consolidated text of 27 July 2026
    Supported claim
    The EU AI Act's human-oversight provision focuses on effective oversight capabilities during use, including monitoring, interpretation, override, intervention, and stopping, rather than using human-in-the-loop as a universal safety label.
    Last verification
    Review due
    Scope limitation
    This describes the cited legal provision for high-risk systems in scope; it does not settle terminology for all system designs.
  3. 3.1Wang et al., Putting Humans in the Natural Language Processing Loop: A Survey
    Source
    Wang et al., Putting Humans in the Natural Language Processing Loop: A Survey
    Source role
    Authoritative source
    Exact locator
    Abstract; pp. 47-52
    Supported claim
    A survey of HITL NLP frameworks describes continuously integrating human feedback to improve a model and reports variation in tasks, goals, human interactions, and feedback-learning methods.
    Last verification
    Review due
    Scope limitation
    This is a survey of HITL NLP research and does not define every operational AI loop or establish accuracy, control, or safety for a particular system.