NIST AI RMF APPLICATION DATA GOVERNANCE INFRASTRUCTURE

Diversity, equity, inclusion, and accessibility

Part of: GV: GOVERN — Policies, Accountability & Culture

Description

Organizational teams building, deploying, and using AI systems reflect diversity, and are proactive in addressing harmful bias and discrimination. AI systems inherit and amplify the biases of their creators — team diversity is a technical control, not just an HR initiative.

Suggested Actions

1

Build AI development teams with diversity across protected attributes, disciplinary backgrounds, and lived experiences — homogeneous teams produce homogeneous blind spots

2

Implement inclusive design practices that engage affected communities in AI requirements, testing, and evaluation — not as an afterthought, but as a design input

3

Establish bias testing protocols throughout the AI lifecycle: training data audits, pre-deployment fairness testing across demographic groups, and ongoing production monitoring

4

Create accessibility requirements for AI system interfaces and outputs — ensure AI-generated content meets WCAG 2.1 AA standards and is usable by people with disabilities

5

Conduct disparate impact analysis before deploying AI systems that affect hiring, lending, insurance, housing, or criminal justice decisions

6

Document and publish demographic performance breakdowns for high-risk AI systems — aggregate performance metrics can mask significant disparities

Related Controls

AI Acceptable Use Policy