Transparency
A core concept in AI ethics referring to the degree to which a system and information about its operation are made public. It supports accountability but is not equivalent to it.
Timeline (last 90 days)
Transparency, explainability, interpretability and accountability are four interrelated but non-interchangeable concepts: transparency refers to the degree to which a system and information about its operation are disclosed, explainability refers to whether understandable reasons can be provided, interpretability refers to how people understand model outputs, and accountability refers to who must answer for decisions and provide redress.
Academia calls the phenomenon in which the decision-making process of recommender systems is difficult to explain the 'algorithmic explainability' problem, and distinguishes two levels: 'transparency' (disclosing how the system works) and 'explainability' (giving human-understandable explanations for specific recommendation results).
All Facts (2)
Transparency, explainability, interpretability and accountability are four interrelated but non-interchangeable concepts: transparency refers to the degree to which a system and information about its operation are disclosed, explainability refers to whether understandable reasons can be provided, interpretability refers to how people understand model outputs, and accountability refers to who must answer for decisions and provide redress.
50%UnverifiedAcademia calls the phenomenon in which the decision-making process of recommender systems is difficult to explain the 'algorithmic explainability' problem, and distinguishes two levels: 'transparency' (disclosing how the system works) and 'explainability' (giving human-understandable explanations for specific recommendation results).
50%