The degree to which a user applies systematic checking practices to AI-generated information, such as source-checking, cross-referencing across independent sources, and contextual appraisal before relying on it. A graded behavioral disposition in handling AI output.
STATE. Operationalized as the proportion of AI-provided claims a user subjects to defined checking steps (independent source confirmation, triangulation, contextual qualification) before use, observed over a set of outputs or via a behavior-frequency self-report. The construct is the level of checking applied.
Proposed measurement protocol (not yet empirically validated): Checking-behavior frequency: per claim, code which of a fixed checklist of verification steps were performed (observed logs or diary); report mean checks-per-claim; if self-reported, use a behavior-frequency scale with test-retest reliability.
https://andreasehstandlicenseofclarityloc.github.io/augmanitai-periodic/#topic-meta-reflection