An inspection result can contain an accurate measurement and still be wrong for the job. Check a part against an obsolete drawing revision and the instrument will record the dimension faithfully, the software will produce a polished report, and the incorrect conclusion will reach everyone who needs it at impressive speed. That scenario is the starting point for a proposal by Gleb Tsipursky, PhD, a behavioral scientist and chief executive of Disaster Avoidance Experts, who argues that manufacturers adopting AI-assisted inspection should attach a version passport to every consequential measurement.
The stakes rise as metrology becomes more connected. An October 26 report on persistent identification described 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 problem, as Tsipursky frames it, is making sure each decision follows the correct version of the requirement, not merely that a number can be traced back to an instrument.
What travels with the measurement
The passport he proposes does not have to be a new software product. It can start as a set of fields that accompany a result:
- 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.
None of these fields is exotic. Most already exist in product data management, inspection planning and laboratory systems, and the ones that matter are those that tie a result to the requirement it was judged against. The proposal is a discipline of keeping them attached, and of refusing to let a conclusion circulate without the requirement behind it. The passport is a convention before it is a purchase.
The scope is deliberately narrow. A passport makes the applicable requirement visible to the next decision-maker; it does not establish whether that requirement was itself the right one. That judgment stays with engineering review and with the people accountable for the design record. What the passport removes is the silent mismatch, the case in which every participant reasons correctly from a definition that no longer governs.
Why AI raises the risk
AI assistants can summarize an inspection report, point to an apparent deviation or recommend which cases need review. A tool that receives results without version context, however, may compare an observation against the wrong specification. The arithmetic can be entirely correct while the conclusion is invalid, and the polished presentation makes the error harder to notice. Tsipursky connects the concern to the National Institute of Standards and Technology's 2026 smart-manufacturing roadmap, which lists integration of heterogeneous data and trustworthy operation among the continuing challenges for industrial AI. A version passport addresses a narrow part of that problem rather than the whole of it.
Testing the idea before buying a platform
Quality teams can evaluate the concept with records they already hold. The suggested protocol is to select a small set of completed inspection records covering three cases: one legitimate out-of-tolerance result, one conforming part and one example with mismatched revisions. An AI assistant is then asked to summarize the evidence, identify the applicable revision and flag missing provenance, and its answers are compared with the determination of an experienced quality engineer.
The evaluation criteria matter more than the headline verdict. Asking how often an assistant labels a part good or bad misses the point; what should be recorded is whether it catches a revision mismatch, whether it invents a tolerance that was never stated, and whether it can point to the approved source behind a conclusion. The mismatched-revision case is the discriminator, because a tool that only restates numbers will summarize it as fluently as the other two while the comparison itself remains invalid.
Digital handoffs make the weakness easy to miss. Each transfer of a record tends to carry the number and drop the conditions under which it was produced, so by the time a decision is made the observation may be separated from the plan and revision that gave it meaning.
Who owns a disputed record
Where records disagree, ownership has to be explicit. 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, in this view, should be settled before an alert appears rather than negotiated during a shipment dispute, when commercial pressure is highest.
Beyond the AI-versus-human comparison
The argument departs from the familiar demand for more comparisons between automated and human inspectors. Even perfect agreement between the two can reinforce the wrong decision if both were working from an obsolete specification, so the reference itself needs scrutiny and not only the agent reading it. Automation makes measurement data travel farther and faster, which is why a reliable chain of custody matters: it keeps the drawing, the feature, the method and the result belonging together at the moment of decision. Traceability of the instrument is not the same as traceability of the requirement, and it is the second that decides whether a part is accepted. On this account, the next generation of connected inspection should be built so that the connection cannot be ignored.