Evaluation
Değerlendirme
D4 · Evaluation, reliability and evidence
AI evaluation is a structured process that produces evidence about how an AI system performs against stated criteria in a stated context.
Review status: 2027-08-28
Technical explanation
It documents qualitative or quantitative performance or assurance criteria under conditions similar to the stated deployment setting.
Conceptual boundaries
Evidence from one setting does not establish how a system will perform in every operational setting.
Provider-neutral example
A team tests whether its support bot answers 100 representative questions correctly.
Limitations
Laboratory measurements can differ from risks that emerge in operational settings.
Related concepts
Atomic claims and evidence
1.1NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0) — D4 evidence slice
- Source
- NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0) — D4 evidence slice
- Source role
- Authoritative source
- Exact locator
- Table 3, MEASURE 2.3, printed p. 29
- Supported claim
- NIST's MEASURE function calls for qualitative or quantitative performance or assurance criteria to be demonstrated for conditions similar to deployment settings and documented.
- Last verification
- Review due
- Scope limitation
- This is risk-management guidance, not a requirement that every evaluation use the same metric or protocol.
2.1NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0) — D4 evidence slice
- Source
- NIST AI 100-1, Artificial Intelligence Risk Management Framework (AI RMF 1.0) — D4 evidence slice
- Source role
- Authoritative source
- Exact locator
- Section 1.2.1, Risk in real-world settings, printed p. 6
- Supported claim
- NIST notes that laboratory measurements can differ from risks emerging in operational settings.
- Last verification
- Review due
- Scope limitation
- This does not make pre-deployment evaluation useless; it limits what it can establish about later use.