THE CAPABILITY CHANGED. DID THE AUTHORITY?

I help organisations determine when AI capability change requires a new decision.

Authority continuity · Revalidation · Execution control · Human–AI decision systems

Same capability. Different conditions. Does the same authority still hold?

BBASISMATERIAL
CHANGE
AUTHORITY
PERMISSION
EXECUTION

Change does not automatically reopen authority.

REVALIDATION?

BASISMATERIAL CHANGE
AUTHORITYSTILL ACTIVE
PERMISSIONSTILL ACTIVE
EXECUTIONIN PROGRESS
CONSEQUENCEINACTIVE

A capability can continue operating after the conditions that justified its authority have materially changed.

AT APPROVALEvidenceAssumptionsScopeUsersWorkflowRisk
APPROVED

A decision for conditions that existed at the time.

What made the old decision current again?

CURRENT STATEModel changedData changedWorkflow changedHuman role changedOperating environment changedConsequence profile changed
PERMISSION STILL ACTIVE

The same capability, in different conditions.

Deployment can establish capability. Continued reliance requires current justification.

AI Authority Continuity Review

A bounded review of one consequential AI-enabled capability to determine whether it is still entitled to do what it does under current conditions.

Explore the Review →
  1. 01ONE CAPABILITY
  2. 02ONE CONSEQUENTIAL WORKFLOW
  3. 03ONE AUTHORITY PATH
  4. 04ONE EXECUTIVE DETERMINATION
BASISWhat currently supports continued reliance?
AUTHORITYWhat decision or mandate still governs?
PERMISSIONWhat is the capability currently allowed to do?
EXECUTIONWhat can actually happen before reassessment changes anything?

BASIS  →  AUTHORITY  →  PERMISSION  →  EXECUTION

Not another maturity score.

The Review produces a decision-relevant determination.

  • CURRENT ALLOWED ACTIONS
  • CURRENT BASIS FOR RELIANCE
  • MATERIAL CHANGES SINCE APPROVAL
  • WHETHER EARLIER AUTHORITY IS BEING RELIED UPON BEYOND THE CONDITIONS THAT SUPPORTED IT

Authority should not silently outrun evidence.

Something important changed. The permission did not.

Not every change requires a new decision. The problem is knowing which one does.

  • MODEL / PROVIDER CHANGED
  • DATA CHANGED
  • WORKFLOW CHANGED
  • HUMAN ROLE CHANGED
  • SCOPE CHANGED
  • CONSEQUENCE PROFILE CHANGED

For organisations where AI can move from output to consequence.

  • AI Governance
  • Enterprise Risk
  • Model Risk
  • Operations
  • Compliance
  • Safety
  • Assurance
  • Regulated environments
  • Human–AI decision environments

OUTPUT → RECOMMENDATION → PERMISSION → ACTION → CONSEQUENCE

Decision Epistemology

Decision Epistemology is an emerging research programme concerned with the epistemic integrity of decision systems over time.

How does a decision system know that the picture it is acting upon still deserves belief?

Working papers · Publications · Research notes · Methods

Explore Decision Epistemology ↗decisionepistemology.org

evidence → picture → confidence → authority → action → consequence

REOPEN?
Wojciech Z. Kaleta

Wojciech Z. Kaleta, PhD
Decision Engineer

More than two decades in maritime aviation.

PhD research in Situational Awareness and safety-critical decision environments.

Human Factors. Human–AI decision systems.

About / research record →

Deliberately bounded.

  • FIXED SCOPE
  • WRITTEN ENGAGEMENT
  • ONE CAPABILITY
  • ONE CONSEQUENTIAL WORKFLOW
  • EVIDENCE BEFORE CONCLUSION

Has the capability changed enough to require a new decision?

If you are responsible for a consequential AI-enabled capability and need to determine whether its current authority still fits current conditions, send a written enquiry.

Fixed-scope · Written-first.

Send a written enquiry →