Digital Lexicon Hub

An international standard establishing a quality model and evaluation criteria tailored to software systems incorporating artificial intelligence.

Structured quality models allow procurement and engineering teams to specify, benchmark, and evaluate AI software quality and reliability against clear contract specifications.

An enterprise procurement team uses the ISO/IEC 25059 quality model to define acceptance testing criteria and SLAs for a commercial computer vision API.

Tailoring quality models to non-deterministic systems requires empirical benchmarking; quality metrics in static test sets may degrade when real-world production data drifts.

Part of the SQuaRE (Systems and software Quality Requirements and Evaluation) series, ISO/IEC 25059 adapts software quality characteristics for AI. It adds AI-specific sub-characteristics such as functional adaptability, user controllability, transparency, robustness, and intervenability, and modifies functional correctness, giving consistent terms for specifying, measuring, and evaluating AI system quality.