Compliance

Human in the Loop (HITL)

Mandatory human review of generative AI outputs before official use.

How the laws define this

All four frameworks agree that human in the loop requires human oversight, control, and review of artificial intelligence systems to ensure accountability and prevent unvetted automated decisions. Under these standards, AI acts as a tool to complement human expertise rather than replace human responsibility for final outcomes. The primary divergence among the sources lies in the specific mechanism and timing of this human involvement. The Seattle Responsible AI Policy sets a strict operational requirement that humans must review Generative AI outputs prior to each official use. Conversely, Towards Responsible AI for All (Approach Document, Part I) - NITI Aayog focuses on ongoing operational controls during execution, such as mechanisms to monitor, override, or halt decisions. Taking a broader approach, the Georgia State AI Governance and Use Guidance and Artificial Intelligence and Georgia's Courts: Final Report of the Judicial Council of Georgia Ad Hoc Committee on Artificial Intelligence and the Courts define the concept as an overarching governance principle or approach ensuring human control and subject matter expert validation.

Synthesised from the 4 statutory definitions below. Each one is quoted in full, with its source.

How jurisdictions differ

What is distinctive in each definition, binding law in force first. Showing 3 of the 4 instruments that define this term; all are listed under Definitions below.

Jurisdiction	Instrument	Status	What is distinctive
Georgia, United States	Georgia State AI Governance and Use Guidance	In Force (Amended)	Frames HITL as a broad principle advocating human control and responsibility for final decisions, positioning AI as a complement to human expertise.
Washington, United States	Seattle Responsible AI Policy	In Force	Limits HITL to human review of Generative AI outputs prior to each use in official City capacity.
Georgia, United States	Recommendations of the Judicial Council of Georgia Ad Hoc Committee on Artificial Intelligence and the Courts	Adopted	Emphasizes HITL as an AI approach with human interaction, oversight, and validation by subject matter experts in critical legal processes.

What it triggers

In Seattle, the Responsible AI Policy describes human review of Generative AI outputs prior to each use in an official City capacity.source
The Recommendations of the Judicial Council of Georgia Ad Hoc Committee on Artificial Intelligence and the Courts guide judicial personnel on generative AI risks, evidence rules, and data security.source

Definitions (4)

A requirement that outputs of Generative AI systems must be reviewed by humans prior to each use in an official City capacity.

Seattle Responsible AI PolicyDefinition 1 of 4

Human-in-the-Loop (HITL) is an artificial intelligence approach that requires human interaction and oversight to ensure the proper functioning and ethical application of AI systems. It emphasizes that AI-generated output should be reviewed and validated by human subject matter experts, especially in critical legal processes.

Artificial Intelligence and Georgia's Courts: Final Report of the Judicial Council of Georgia Ad Hoc Committee on Artificial Intelligence and the CourtsDefinition 2 of 4

Human-in-the-Loop (HITL) is a principle advocating for human control over AI systems and human responsibility for final decisions made with AI support. It ensures AI acts as a tool to complement human expertise, not replace it.

Georgia State AI Governance and Use GuidanceDefinition 3 of 4

A governance and design principle requiring meaningful human oversight, intervention, or escalation mechanisms during an AI system's operation so humans can monitor, override, or halt automated decisions where appropriate. The document frames human-in-the-loop/on-the-loop controls as essential for accountability, responsible procurement, and mitigation of harms in public-sector deployments.

Towards Responsible AI for All (Approach Document, Part I) - NITI AayogDefinition 4 of 4

Not to be confused with

Human involvement

Human involvement refers to the degree of human review, approval, or override of AI outputs, while HITL specifically requires mandatory human oversight and interaction in AI processes.

Responsible Experimentation

Responsible experimentation focuses on controlled, risk-aware testing and use of AI tools by staff, whereas HITL centers on human oversight of AI outputs and decisions.

Questions and answers

What does Human in the Loop mean in Georgia state government AI guidance?

In the Georgia State AI Governance and Use Guidance, HITL is a principle for human control over AI systems and responsibility for final decisions. It positions AI as a tool complementing human expertise. The guidance establishes security and ethical standards for executive branch agencies.

Is human review required for AI in Seattle city services?

Seattle's Responsible AI Policy describes human review of Generative AI outputs prior to each use in official City capacity. It guides ethical AI use focusing on equity, transparency, and risk mitigation.