Significant judgment
Every significant judgment is identified before consequence and classified by how AI and the decision-maker formed it.
AI JUDGMENT & CONTROLTHESIS III · SEPTEMBER 2026
Control begins before consequence.
Once AI contributes substantively to judgment, assurance over the model, output or act of review is no longer sufficient. This thesis makes significant judgment the control object, places detect control before prevent control and uses the implementation boundary to distinguish control over judgment from the separate work required after consequence.
Detect control
Detection establishes the judgment configuration, retained independence and fidelity of purpose, evidence and context.
Prevent control
The organization determines whether the detected form may proceed, requires restricted AI participation or prevents implementation.
SIFTER™ architecture
Source, Independence, Fidelity, Threshold, Exposure and Response connect formation analysis to control determination.
RIGHTS
This publication contains the proprietary SIFTER™ Judgment Control Architecture and associated Teraxys analytical frameworks. See the final PDF for the complete proprietary-frameworks and rights notice governing citation, reproduction, adaptation, implementation and commercial use.
AI JUDGMENT & CONTROL