Robustness
Sağlamlık
D4 · Evaluation, reliability and evidence
Robustness is the goal of appropriate system functionality across a broad stated set of conditions, including unanticipated uses.
Review status: 2027-08-28
Technical explanation
It assesses appropriate functionality across the stated conditions and circumstances.
Conceptual boundaries
Robustness is not universal resilience; it is evaluated in relation to stated conditions and circumstances.
Provider-neutral example
A team checks whether a document classifier still works on blurry scans.
Limitations
Evidence from selected conditions does not establish universal resilience outside them.
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. 14
- Supported claim
- NIST describes robustness as appropriate system functionality in a broad set of conditions and circumstances, including uses not initially anticipated.
- Last verification
- Review due
- Scope limitation
- A broad set of conditions is still bounded by the system and assessment context; it is not a claim of universal resilience.
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 says that accuracy and robustness contribute to validity and trustworthiness and can be in tension in AI systems.
- Last verification
- Review due
- Scope limitation
- This identifies a possible trade-off; it does not say that every accuracy improvement reduces robustness.