Manufacturers adopting AI-assisted inspection should attach a "version passport" to every consequential measurement, argues behavioral scientist Gleb Tsipursky. The proposal targets a failure mode in which a measurement is numerically accurate yet still wrong for the job because it was checked against an obsolete drawing revision.
In the scenario Tsipursky describes, a part is inspected against a superseded drawing. The instrument records the dimension faithfully and the software produces a polished report, but the conclusion reaching decision-makers is incorrect, delivered with impressive speed.
Why version context matters
The argument builds on connected metrology. Tsipursky cites an October 26 Metrology News account of persistent identification, which describes how Quality Information Framework data can preserve the identity of a measured characteristic across product definitions, inspection plans and results. The technical promise is traceability; the management challenge, he writes, is ensuring each decision follows the correct version of the requirement.
AI tools can summarize an inspection report, flag an apparent deviation or recommend which cases need review. But a system that receives results without version context may compare an observation against the wrong specification. The arithmetic can be correct while the conclusion is invalid.
Tsipursky points to the National Institute of Standards and Technology's 2026 smart-manufacturing roadmap, which highlights integration of heterogeneous data and trustworthy operation as continuing challenges for industrial AI. A version passport, he argues, targets a specific slice of that problem by making a measurement's origin and applicable conditions visible to the next decision-maker.
What the passport contains
The passport need not be a new software product, according to the commentary. It can start as a set of fields travelling with a measurement:
- the approved drawing or model revision;
- the characteristic identifier;
- the inspection-plan revision;
- the instrument or method used;
- the result timestamp;
- the identity of the person or system that approved the disposition.
A small test before platform integration
Quality teams can try the idea before committing to a new platform, Tsipursky suggests. He recommends selecting a small set of completed inspection records, including one legitimate out-of-tolerance result, one conforming part and one example with mismatched revisions, then asking an AI assistant to summarize the evidence. The assistant should be required to identify the applicable revision and flag missing provenance, with its responses compared against an experienced quality engineer's determination.
Rather than measuring only how often the assistant labels a part good or bad, he advises recording whether it catches a revision mismatch, whether it invents an unstated tolerance and whether it can point to the approved source behind a conclusion. Those, he says, are the errors digital handoffs often conceal.
Ownership when records disagree
The column also calls for clear ownership. A quality engineer may determine whether a measurement is valid, design engineering confirms the governing definition, and manufacturing decides whether production can continue. The escalation route should be settled before an alert appears rather than negotiated during a shipment dispute.
Tsipursky distinguishes his argument from the familiar demand for more AI-versus-human comparisons. Even perfect agreement between human and automated inspectors can reinforce the wrong decision if both were given an obsolete specification, he writes, so the reference itself needs scrutiny. Automation lets measurement data travel farther and faster, and a reliable chain of custody ensures the drawing, feature, method and result still belong together when the decision is made.
Tsipursky is CEO of Disaster Avoidance Experts and author of The Psychology of AI Adoption at Work: From Resistance to Results (Georgetown University Press, 2026).