Skip to main content

COMPEL Glossary / GL-55

Bias Delta

The measured difference in a model's performance or outcome distribution across protected groups, expressed against a defined fairness threshold.

What this means in practice

Bias delta is computed per protected attribute (for example gender, age band, geography) using disparate-impact ratios, equal-opportunity gaps, or calibration differences, and is monitored over time so that drift into unfair behavior is detected before it reaches users.

Context in the COMPEL framework

A core metric of the Responsibility dimension. Captured in the Evaluate stage and reviewed in Learn; feeds fairness gates in release governance.

Where you see this

Bias Delta is most commonly referenced when teams work across the Evaluate and Learn stages — especially within the Agent Governance layer . It appears in governance artifacts, assessment instruments, and delivery playbooks wherever COMPEL is operationalized.

Related COMPEL stages

Related domains

Synonyms

fairness gap , disparate impact delta , protected-group delta

See also

  • Trust & Performance Dimensions — The eight continuous-measurement axes against which every AI transformation is evaluated in COMPEL: Value, Reliability, Safety, Responsibility, Compliance, Security, Sustainability, and Adoption.
  • Responsibility Index — A composite scorecard metric for the Responsibility dimension, combining bias delta, explainability coverage, and human-oversight effectiveness into a single index used in executive reviews.