In AI governance, there is an increasing need to assess whether an AI system remains adequately trustworthy over time, whether observed changes are tolerable, and how these judgments can be documented transparently and contestably. Existing work on AI trustworthiness tends to be either too high-level to support lifecycle monitoring and reassessment or too narrowly focused on metrics, failing to connect with governance needs. Therefore, we propose a lightweight methodology for auditable trust levels in AI governance.
This methodology consists of two components: a formal framework for representing and learning trust levels, and a lightweight AI lifecycle governance procedure for documenting, monitoring, and reassessing these levels. The formal framework models governance-relative trustworthiness through a context-sensitive protocol of measurable dimensions, learning trust levels as interpretable rules over trust profiles. Using decision trees as an interpretable proof-of-concept model class, the methodology yields explicit trust plateaus, readable level transitions, and two simple lifecycle diagnostics: boundary margins and profile drift.
The governance procedure embeds these formal objects in a conformity-oriented workflow for design-time labeling, post-deployment monitoring, reassessment, and reporting. It also assigns human responsibilities and control gates for protocol design, validation, monitoring, and reassessment. We illustrate the methodology on synthetic AI lifecycle traces involving degradation, shocks, updates, heterogeneous monitoring cadences, and system comparison. Our methodology does not replace legal or expert judgment but supports conformity documentation and lifecycle monitoring by providing an evidential basis for documenting and tracking AI governance-relevant changes over time.
Blogger's Review: The proposed methodology for auditable trust levels holds significant importance in AI governance, providing a systematic framework for monitoring and managing trustworthiness. Utilizing interpretable models like decision trees enhances the transparency and traceability of governance processes, marking a valuable exploration in the realm of AI trust governance.