Quality Measures and Assurance for AI (Artificial Intelligence) Software

摘要

This report is concerned with the application of software quality and evaluation measures to AI software and, more broadly, with the question of quality assurance for AI software. Considered are not only the metrics that attempt to measure some aspect of software quality, but also the methodologies and techniques (such as systematic testing) that attempt to improve some dimension of quality, without necessarily quantifying the extent of the improvement. The report is divided into three parts Part 1 reviews existing software quality measures, i.e., those that have been developed for, and applied to, conventional software. Part 2 considers the characteristics of AI software, the applicability and potential utility of measures and techniques identified in the first part, and reviews those few methods developed specifically for AI software. Part 3 presents an assessment and recommendations for the further exploration of this important area.