What is Autonomy Appetite?

Define how much authority an AI-enabled decision may receive, and what evidence must support that delegation.

Autonomy Appetite connects AI transformation goals to bounded decision authority. Learn how AIDIVE separates what AI may do from how humans supervise it.

Start with the decision

Autonomy Appetite is the maximum authority an organisation is prepared to delegate for a decision class under current evidence and control conditions. It gives leaders a way to discuss delegation through the consequences of a decision, rather than through a model's technical capability.

A useful assessment considers impact, reversibility, uncertainty, frequency and the ability to detect harm. The result should state the permitted scope, the accountable owner and the conditions that require escalation, restriction or a stop.

Choose participation and oversight separately

AIDIVE connects Assist, Recommend, Act and Escalate-Stop to delegated decisions. Human involvement is a separate design choice: human-in-the-loop, human-on-the-loop or human-out-of-the-loop. Neither set is a universal maturity ladder. The appropriate configuration depends on the decision and operating context.

Human presence alone does not establish effective oversight. The person must have the context, time, competence and authority to intervene. AIDIVE's V1 guidance reserves HOOTL for low-risk, reversible and observable decision classes, with named human accountability, escalation, auditability and technically available takeover or stop capability.

Illustrative example: a service credit

A service team could allow an agent to draft a credit recommendation while a person authorises the commitment. A later, bounded delegation might permit action only for an approved decision class, using current policy and defined limits. Missing evidence, conflicting policy or an exception would trigger escalation.

Before changing that authority, test normal, boundary, conflict and degraded-data cases with the people expected to intervene. Record the result and review it when the model, tools, context or consequences change.

• Name the decision class and its accountable human owner.
• Specify permitted actions, oversight, thresholds and safe states.
• Demonstrate intervention and record the evidence required for any change.

Take this into your next conversation.

Delegation expands through explicit decisions and evidence. Technical capability does not grant authority.

Framework source

Editorial adaptation from the supplied AIDIVE Version 1 master documents.

• AIDIVE Alignment Dimension Version-1.docx · A.2.2 How decisions are made with AI; A.2.3 Autonomy Appetite: From Strategic Posture to Decision Mode
• AIDIVE Alignment - Practices & Playbook Version-1.docx · AR-A06 Delegated Governance & Decision Mode Design
• AIDIVE Ethics Dimension Version-1.docx · E.5.1 Autonomy classification and delegated-authority envelope; E.5.4 Human oversight, approvals and intervention effectiveness

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.