Human-in-the-Loop
What is Human-in-the-Loop?
An AI system design pattern where human oversight is integrated into the decision-making process. Human-in-the-loop ensures that critical decisions receive human review and is required for high-risk AI systems under the EU AI Act.
Terms that appear alongside human-in-the-loop
Each of these is named in at least one of the same controls as human-in-the-loop. The number is how many controls name both.
- nist 3 shared controls
- eu ai act 3 shared controls
- impact assessment 2 shared controls
- audit 2 shared controls
- gdpr 2 shared controls
- risk assessment 2 shared controls
- incident response 2 shared controls
- remediation 2 shared controls
Frameworks that govern human-in-the-loop
What the standards actually require on human-in-the-loop
Requirements naming human-in-the-loop across 6 standards, quoted from the control text.
AI provides recommendations while a human makes the final decision and is accountable
AIGF-2.1 · Human-in-the-Loop →Requirement defined in ISO/IEC TR 24028:2020, clause 9.4.2 (Human-in-the-loop control points). See licensed source for normative text.
iso-iec-tr-24028-2020::9.4.2 · Human-in-the-loop control points →Human Oversight is mandated by the Human-Centric Principle (1st of 10 Principles) per Japan AI Guidelines for Business + reinforced by APPI 2022 Amendment Article 21-2 right to human review + intersects with Hiroshima AI Process Code of Conduct.
JP-AIG-Human-Oversight-Control-In-the-Loop-On-the-Loop-Article-22-GDPR-Equivalent-Automated-Decision-Restrictions · Japan AI Guidelines Human Oversight + Human-in-the-Loop + Human-on-the-Loop + Human-out-of-Loop + Article 22 GDPR Equivalent APPI Automated Decision Restrictions + Override Capability + Pause Functionality + Audit Trail →Based on the risk assessment, determine the category of human involvement, human-in-the-loop, human-over-the-loop, or human-out-of-the-loop, commensurate with the assessed risk.
AIGE-HI-3 · Determine the level of human involvement →Clauses 5.2 + 5.3 establish stakeholder identification and engagement. Per public IEEE 7000-2021 abstract + Wikipedia + academic literature (full IEEE text NOT reproduced): stakeholder identification including: direct users (purchasers + operators);
IEEE7000-Stakeholders-DirectIndirectMarginalised-Engagement-CoDesign · IEEE 7000 Clauses 5.2 + 5.3 - Stakeholder Identification + Engagement + Direct + Indirect + Marginalised + Co-Design + Human Oversight →Apply MANAGE function including: content provenance implementation per NIST AI 600-1 MGT-1 (cryptographic watermarking + C2PA Content Credentials + media authentication) + human oversight integration per MGT-2 (human-in-the-loop + human-on-the-loop + human-out...
NISTAI600-5 · Manage - Content Provenance, Human Oversight, and Incident Response →Questions people ask about human-in-the-loop
What is Human-in-the-Loop?
Why is Human-in-the-Loop important for compliance?
Which compliance frameworks address Human-in-the-Loop?
Where can I learn more about Human-in-the-Loop?
See how Human-in-the-Loop applies across compliance frameworks
Our platform maps 686 frameworks with 311K cross-framework control mappings. Explore how this concept is addressed across standards.