AI Authority and Human Control

Treat Assist, Recommend and Act separately from HITL, HOTL and bounded HOOTL at each material decision or action seam.

In AIDIVE, AI authority and human oversight are separate operating choices. Design both around the decision class, with named human accountability and working intervention.

Why it matters

Authority states what AI may do. Human control describes how a person approves or oversees that contribution. These choices need independent, explicit design.

Define the permitted AI contribution

Assist, Recommend and Act describe the authority available at a decision or action seam. Authority stays scoped to the task and operating context; technical access does not grant business authority. Some decisions remain prohibited, escalation-only or non-delegable.

Position human control

HITL, HOTL and bounded HOOTL describe how human decision or oversight is positioned around that authority. Different seams in one workflow can use different configurations. They are separate design properties, not maturity levels or a universal autonomy ladder.

Keep autonomy bounded by evidence

Apply the current decision rights, risk classification and Autonomy Appetite. Bounded HOOTL is reserved for eligible low-risk, reversible and observable decision classes with a named accountable human and tested intervention or safe-stop capability.

• Specify intervention conditions, timeout and safe default.
• Give receiving humans context, time and usable authority.
• Route material authority or control changes through the approved evidence and decision path.

Evidence to make visible

• An authority mode and human-control configuration per material seam
• Named human accountability and intervention capacity
• A bounded action envelope and tested safe state

Chapter map

• 2.7.7 HITL / HOTL / Bounded HOOTL and AI Authority Boundaries
• AI Authority, Human Control and Escalation Design
• Override, Fallback and Safe Intervention Mechanisms
• Decision Evidence, Logging and Auditability

Framework source

Editorial adaptation from the supplied AIDIVE Version 1 master documents.

• AIDIVE - Architecture & Rhythm - Agentic AI Operating Model Version-1.docx — §§2.7.6–2.7.7; §2.6.4. AIDIVE Architecture & Rhythm - Maturity Version-1.docx — §1.1 Phase 3.

Go a little deeper.

Put AI transformation
into practice.

Build capability through training, contribute your expertise or explore a connected workspace for applying the holistic AIDIVE framework.

Build the capability.

Develop the skills to connect AI transformation strategy, business outcomes, people and governance to your own decisions and responsibilities.

Grow the practice.

Bring your learning, delivery or research expertise to the holistic AIDIVE approach to enterprise AI transformation.

Bring AIDIVE into your work.

See how the proposed workspace could connect AI transformation learning, maturity evidence and practical artifacts.

There is more to explore.