Evaluating Maintenance AI Against the Work It Must Support
Executive summary
The central problem in maintenance AI evaluation is not a shortage of technology. It is that offline accuracy does not represent missing data, unusual configurations, time pressure, automation bias, or downstream operational consequence. A useful design must preserve operational meaning while making the next decision easier to inspect.
This paper proposes a bounded approach: combine scenario testing, prospective shadow use, human-factors review, cohort metrics, and outcome adjudication. The intent is decision support with explicit evidence and accountable authority—not an automated substitute for approved maintenance data, engineering judgment, or licensed action.
Evaluating Maintenance AI Against the Work It Must Support
Where do governed evidence, controls, accountable owners, and release authority sit?
1. Define the operational decision
Programs often begin by collecting available data or selecting a platform. That reverses the useful order. The team should first identify who must decide, when the decision occurs, which evidence is authoritative, what uncertainty is acceptable, and which action remains under qualified control.
For maintenance AI evaluation, the dominant constraint is that offline accuracy does not represent missing data, unusual configurations, time pressure, automation bias, or downstream operational consequence. The product boundary should therefore be written as a decision contract: inputs, freshness, effectivity, interpretation rules, exclusions, reviewer role, downstream record, and measurable outcome. This contract gives engineering and operations a shared definition of done.
Evaluating Maintenance AI Against the Work It Must Support
Which requirement, evidence, owner, control, and review status must remain traceable?
2. Preserve evidence before interpretation
Source records should retain identity, event time, ingestion time, configuration context, revision, lineage, and quality state. Normalized concepts are valuable, but they should never overwrite what the source actually reported. Investigators need to reproduce the view that existed when a decision was made.
The recommended design is to combine scenario testing, prospective shadow use, human-factors review, cohort metrics, and outcome adjudication. Derived features, rules, statistical output, retrieved text, and generated synthesis should be distinguishable in storage and in the user interface. That separation supports correction without rewriting history and allows reviewers to challenge an inference while accepting the underlying evidence.
Evaluating Maintenance AI Against the Work It Must Support
How do assurance evidence, review, restriction, incident response, and corrective action connect?
3. Engineer the authority boundary
Operational software can assemble context, identify patterns, rank attention, and prepare a structured brief. It cannot create maintenance authority. The interface must identify the governing source, effective revision, responsible role, and required disposition. Override and abstention are normal system behaviors.
The most important anti-pattern is promoting a model from a random historical split without testing the future workflow. It tends to appear efficient because ambiguity disappears from the screen. In reality the ambiguity has only been hidden from the person accountable for the decision. Controls should make missing context, conflict, and inapplicability prominent enough to change behavior.
Evaluating Maintenance AI Against the Work It Must Support
Which evidence permits release, restriction, rollback, or withdrawal?
4. Implementation, governance, and limitations
A credible first release should write evaluation cases with maintenance experts and predefine release, restriction, and rollback criteria. The team should conduct prospective shadow use, compare product output with actual engineering reconstruction, and record why reviewers accept, modify, or reject the result. Expansion should depend on evidence quality and workflow value rather than demonstration appeal.
Governance belongs in the service itself: access control, source eligibility, versioning, release evidence, monitoring, rollback, retention, and outcome stewardship. Limitations should be published by fleet, configuration, operating regime, source availability, and decision type. When applicability cannot be established, the safe result is a visible abstention.
Measures should connect technical behavior to the decision contract. Useful families include evidence completeness, freshness, unresolved identity, reviewer correction, false escalation, missed significant cases, decision latency, recurrence, and outcome-linkage quality. These measures are meaningful only when segmented by the operational conditions that influence them.
Key takeaways
- Begin with a named decision, accountable role, and evidence contract.
- Preserve recorded facts separately from normalization and inference.
- Design explicitly against promoting a model from a random historical split without testing the future workflow.
- Write evaluation cases with maintenance experts and predefine release, restriction, and rollback criteria.