Definition
Measures the degree to which the AI customer risk tiering model clusters customers in medium-risk categories relative to the expected distribution, detecting optimization for efficiency at the expense of accuracy.
Formula
RTCI = (High_risk_AI% - High_risk_reference%) / High_risk_reference%; companion TMA = Downgrades / Upgrades
Proposed formula, pending validation pilots.
At a Glance
Category
AI Fraud & Financial Crime Calibration
Decision Supported
Is the AI burying genuinely high-risk customers in a medium-risk bucket to avoid expensive enhanced due diligence?