Transparency as Design Publicity: Explaining and Justifying Inscrutable Algorithms

Michele Loi, Andrea Ferrario

Abstract: This paper argues that transparency for machine-learning systems should not be reduced to post-hoc interpretability. It proposes “design publicity”: explaining an algorithm as an intentional artifact designed for specific goals, with evidence about how well those goals are achieved. The account links transparency to justification, contestability, and procedural fairness.