A systematic error that favours or disadvantages some people or outcomes. NIST sorts AI bias into three categories: systemic, statistical and computational, and human 1. Bias can enter at any lifecycle stage, so it is tested per stage rather than once.
- Developed in
- ch. 16, Where bias enters the lifecycle
- Chapters
- ch. 16, Fairness & XAI
- Contrast with
- Fairness
- Source
- 1 numbered reference, listed below
Where it is used
18 chapters of the Body of Knowledge use the term. Each link opens the first section that does.
- 02 · Why Now What changes when governance is engineered 1 mention
- 04 · The Stack Data governance across the stack 6 mentions
- 05 · Patterns Pattern: Dataset Admission Gate 1 mention
- 08 · Regulatory Map EU AI Act, post-Omnibus 10 mentions
- 10 · Reading List Government and regulator guidance beyond the EU 6 mentions
- 11 · AI Defined Kinds of AI that change the governance problem 4 mentions
- 12 · Governance Program AI literacy as code 4 mentions
- 13 · Risk Management Identifying risk: sources, factors and stakeholders 2 mentions
- 14 · Development The use-case record 11 mentions
- 15 · Deployment Operating the system 2 mentions
- 16 · Fairness & XAI Opening 27 mentions
- 17 · Incidents Root-cause analysis 2 mentions
- 18 · EU AI Act The Act and the Omnibus 8 mentions
- 19 · Privacy & AI Controller duties across the AI supply chain 6 mentions
- 20 · Existing Law Non-discrimination 7 mentions
- 21 · AI Laws Worldwide United States: state laws that bind private organisations 4 mentions
- 22 · Principles & Standards The instruments at a glance 10 mentions
- 23 · AI Agents EU AI Act hooks for agents 1 mention
Patterns that use this term
4 pattern pages use the term, most mentions first.
- Dataset Admission Gate 8 mentions
- Fairness Eval Suite 5 mentions
- Drift & Fairness Monitor 3 mentions
- FRIA-as-Code 1 mention
Related terms
Sources
- [1] NIST SP 1270, Towards a Standard for Identifying and Managing Bias in Artificial Intelligence (three categories of AI bias: systemic, statistical and computational, and human). NIST. 2022-03-15. https://doi.org/10.6028/NIST.SP.1270 (verified: primary)
Definitions of legal terms paraphrase the cited text, which governs. Dated statements are as of .