CKO-015Validation & EvaluationStrong evidence
What is stability?
Stability is the extent to which an AI system produces consistent outputs when given the same input repeatedly.
In more detail
Many generative AI systems are probabilistic and may generate different responses to the same prompt. Stability assessment examines whether those differences are trivial wording changes or meaningful differences that affect decisions. Stability is particularly important for evidence synthesis tasks where reproducibility matters.
Why it matters
Inconsistent outputs may undermine confidence in AI-assisted workflows.
Decision rule
Assess stability before relying on AI for important decisions.
Common misconception
“A tool is reliable if it gives a good answer once.”
At a glance
- Evidence strength
- Strong
Related concepts
Key takeaway
Reliable tools should behave consistently under consistent conditions.