Transparency Explainable AI
noun
Definition 1: A property of an interpretable model or system whereby information about its internal operation and the grounds of its outputs is accessible to humans [Bakanach 2023/2024].
Example in context: “By exploring how explainability in recommender systems can strengthen transparency and foster trust, the study addresses a central concern for both the scientific community and the public.” [Govea et al. 2024]
Definition 2: Organizational openness in the processes of AI development, reporting, and evaluation, including clear disclosure of limitations, assumptions, and risks [Regulation (EU) 2024/1689].
Example in context: “To build towards more responsible use of health data, we firstly advocate for transparency – thoughtful, self-critical assessment of datasets, and clear reporting of limitations and biases, enabling a better understanding of whether a dataset serves as a sound basis from which to draw scientific conclusions or to build a technology.” [Alderman et al. 2024]
Related terms: interpretability, explainability, accountability, disclosure, reporting transparency, trustworthy AI