Skip to main content

COMPEL Glossary / demographic-parity

Demographic Parity

Demographic parity is a mathematical fairness criterion requiring that an AI system's positive outcomes (such as loan approvals, job interview invitations, or benefit eligibility) are distributed equally across different demographic groups, regardless of the group's representation in the underlying data.

What this means in practice

While intuitive, demographic parity can conflict with other fairness metrics and may not be appropriate in all contexts, making it one of several fairness criteria that organizations must choose between based on their specific situation and values. For organizations, understanding that multiple, sometimes incompatible definitions of fairness exist is essential for making informed governance decisions. In COMPEL, demographic parity is one of several fairness metrics discussed in the ethical framework of Module 3.4, Article 4 on advanced ethics architecture.

Why it matters

Understanding that multiple, sometimes incompatible definitions of fairness exist is essential for making informed AI governance decisions. Demographic parity is intuitive but can conflict with other fairness metrics, and blind application can produce outcomes that stakeholders find unjust. Organizations must make deliberate, documented choices about which fairness criteria apply to each AI system rather than assuming a single definition suffices.

How COMPEL uses it

Demographic parity is one of several fairness metrics discussed within the ethical framework of Module 3.4, Article 4 on advanced ethics architecture. During Model, COMPEL guides organizations to select appropriate fairness metrics based on context, values, and legal requirements. The Evaluate stage measures fairness outcomes against chosen metrics, and discrepancies trigger review through the AI Ethics Board governance structure.

Related Terms

Other glossary terms mentioned in this entry's definition and context.