Reliability
Güvenilirlik
D4 · Evaluation, reliability and evidence
For an AI system, reliability is the goal of overall correct operation under expected-use conditions over a stated period.
Review status: 2027-08-28
Technical explanation
It is assessed through evidence of intended performance under monitored conditions over time.
Conceptual boundaries
Reliability concerns overall correct operation under expected-use conditions over time; it is not interchangeable with validity, accuracy, robustness, or safety.
Provider-neutral example
A helpdesk checks whether its routing assistant keeps working accurately each week.
Limitations
Ongoing testing and monitoring provide evidence under monitored conditions; they do not guarantee the absence of future failure.
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
- Section 3.1, printed p. 13
- Supported claim
- NIST describes reliability for AI systems as a goal of overall correct operation under expected-use conditions over a given period, including the system's lifetime.
- Last verification
- Review due
- Scope limitation
- This is NIST's AI-system description and does not reduce reliability to a single model score.
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 3.1, printed p. 14
- Supported claim
- NIST states that validity and reliability for deployed AI systems are often assessed through ongoing testing or monitoring that confirms intended performance.
- Last verification
- Review due
- Scope limitation
- Ongoing testing can supply evidence under the monitored conditions; it does not guarantee absence of future failure.
3.1NIST 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 3.1, printed p. 14
- Supported claim
- NIST states that accuracy and robustness contribute to validity and trustworthiness, while validity and reliability for deployed systems are assessed through ongoing testing or monitoring.
- Last verification
- Review due
- Scope limitation
- These relationships do not make accuracy, robustness, validity, and reliability interchangeable.
4.1NIST 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
- Sections 3.1-3.2, printed pp. 13-14
- Supported claim
- NIST lists Valid & Reliable and Safe as distinct trustworthiness characteristics and states that accuracy and robustness contribute to validity and trustworthiness.
- Last verification
- Review due
- Scope limitation
- These characteristics interact but are not interchangeable.