Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yu, Jayne, YU, Jieyin
Format: Recurso digital
Sprache:
Veröffentlicht: Zenodo 2025
Online-Zugang:https://doi.org/10.5281/zenodo.16566359
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Inhaltsangabe:
  • <p>We propose the Synchrony Rate (SR) engine, a framework for quantifiable constitutional AI accountability through multi-dimensional persona validation. By embedding governance directly into AI decision-making via runtime constitutional assessment, SR advances beyond binary compliance toward continuous, verifiable alignment.<br>Our proof-of-concept achieves 87.5% decision accuracy (175/200 cases) in a 200-case evaluation, achieving 100% in governance-critical domains such as ethics, precedent, and edge-case reasoning. The SR framework operationalizes constitutional programming through five-component assessment (PHS, SYM, VAL, EMO, FX) and dual consensus mechanisms (Collapse/CoCollapse) for democratic multi-agent coordination. This work establishes a mathematical foundation for constitutional AI, addressing the urgent need for quantifiable governance in emerging regulatory landscapes.</p>