Reliability

A metric measuring an AI system's ability to maintain performance and stability under varied, unseen, or adversarial conditions.

A metric that measures the proportion of correct predictions or outputs produced by a model or AI system out of all evaluated cases.

A metric that measures whether an AI system's output is factually accurate, logically valid, and satisfies the intended task or expected result.

A metric that measures how reliably an AI system produces stable, coherent, and similar outputs for equivalent inputs or repeated evaluations.

The proportion of attempted tasks or requests that an AI agent or system completes successfully.