Measuring Decision-Record Reconstructability in Intelligent Systems: A Six-Field Multi-Criteria Evaluation Instrument for Inclusive AI Governance

This record archives the preprint version of the manuscript "Measuring Decision-Record Reconstructability in Intelligent Systems: A Six-Field Multi-Criteria Evaluation Instrument for Inclusive AI Governance". The manuscript was submitted to the International Journal of Computational Intelligence Systems for the Special Issue / Collection "Intelligent Systems for Sustainable and Inclusive Futures" as a Research article. The paper addresses a downstream gap in intelligent systems (IS) and decision support systems (DSS): whether records produced when a human operator acts on an IS recommendation preserve enough informational structure for post-event accountability, independent audit, contestation, and inclusive governance. It introduces the Boundary Transparency Index (BTI), a six-dimensional 0-12 multi-criteria evaluation instrument for decision-record reconstructability, and the Decision-Event Record (DER), a companion structured logging protocol for human-IS commitment interfaces. The validation study applies BTI to sixty California DMV autonomous vehicle disengagement records using two independent blinded v2-format raters and one supplementary cross-format rater. A supplementary audit of 46 high-stakes UK public-sector AI transparency records is used to test whether the reconstructability gap appears in real public-sector IS deployments. This is a preprint / submitted manuscript version. It has not yet been peer reviewed or formally accepted for publication. The version of record, if accepted, will appear in the journal. The manuscript is part of a broader research programme on decision-event reconstructability, BTI, DER, and AI-mediated accountability. This preprint is linked to the programme Root Index and associated DOI-indexed evidence records to support traceability of the instrument, validation materials, and supplementary audit evidence.