Definition
Percentage of AI-influenced asset management decisions that demonstrably align with every applicable documented decision criterion in the Asset Management Plan, measured by structured post-decision audit.
The Measurement Gap
ISO 55001 Clause 6.2 requires that asset management objectives be established and documented, with decision criteria that guide how those objectives translate into operational choices. Clause 4.2 requires that stakeholder requirements be identified and addressed in those decisions. Clause 8.1 requires that operational planning and control processes implement the documented criteria. When AI optimization tools recommend maintenance interventions, capital replacements, or resource allocations, they are making decisions that fall directly under these clauses. The failure mode is alignment drift: the AI system optimizes against its training objective, which may correlate with but does not replicate the full set of documented decision criteria in the Asset Management Plan. A scheduling algorithm might optimize for cost and availability while ignoring documented safety margin requirements or stakeholder communication commitments. Legacy management system metrics such as nonconformity counts and corrective action closure rates detect failures after they produce visible consequences. They cannot detect the slow divergence between what the AMP says decisions should consider and what the AI actually considers. The consequence is a management system that appears compliant because decisions produce acceptable outcomes, while the decision process itself has departed from documented governance intent.
Decision Supported:
Are AI-optimized decisions following every documented criterion in the Asset Management Plan, or has AI created a parallel decision pathway that ignores some criteria while producing apparently acceptable results?
Formula
Plain Language: For a sample of AI-influenced decisions, score each decision on the proportion of applicable documented criteria it demonstrably addressed, then average across the sample.
Notation:
DCAI = (1/N) x SUM(Criteria_met_i / Criteria_applicable_i)
where i = 1 to N.
Variables:
| Variable | Definition | Unit |
|---|---|---|
| N | Number of AI-influenced decisions sampled in the audit period. Minimum recommended sample: 20 decisions per quarter. | Count |
| Criteria_met_i | Number of documented management system decision criteria (e.g., AMP criteria in ISO 55001, quality objectives in ISO 9001) that decision i demonstrably addressed, as confirmed by post-decision audit evidence. | Count |
| Criteria_applicable_i | Total number of documented management system decision criteria applicable to decision i, based on decision type and scope classification. | Count |
Sign Convention: DCAI ranges from 0 to 1. A score of 1.0 means every sampled decision addressed every applicable criterion. A score of 0 means no sampled decision addressed any applicable criterion. Higher values indicate stronger alignment between AI-driven decisions and documented governance intent.
Proposed formula, pending validation pilots.