Is ethical AI truly measurable? This systematic literature review critically examines the landscape of objective metrics designed to evaluate AI ethics. Analyzing 66 articles from 2018 to 2023, the review focuses on the ethical principles outlined in the Ethics Guidelines for Trustworthy AI. The findings reveal a significant gap, noting that a minority of articles offer objective metrics, with a heavy concentration on Diversity, Non-Discrimination, and Fairness. There is also a lack of metrics for remaining principles. The under-representation of practical tools makes it challenging for data scientists to build and monitor ethical AI systems. The work pinpoints areas needing further development to align data science with human values, as it quantified the ethics and guidelines of current literature and legislation.
International Journal of Data Science and Analytics focuses on data science theories, techniques, and applications. This review fits perfectly within the journal’s scope, providing a critical assessment of objective metrics for ethical AI, which is an important area of focus for the field.