Explanation and Agency: exploring the normative-epistemic landscape of the “Right to Explanation”

Abstract A large part of the explainable AI literature focuses on what explanations are in general, algorithmic explainability is more specifically, and how to code these principles into systems. Much less attention has been devoted question why decisions systems should be whether there ought a right explanation why. We therefore explore normative landscape need for individuals having such explanation. This exploration particularly relevant medical domain where (im)possibility high both research practitioners’ agenda. The dominant intuition overall that play key role health context. Notwithstanding strong explanation, intuitions can wrong. So, we than an appeal when it comes explaining significance being subject AI-based decision-making. aim paper provide account might underlie intuition. defend ‘symmetry thesis’ according which no special reason have ‘machines’ broad sense, make decisions, recommend treatment, discover tumors, so on. Instead, argue cases involve automated processing significantly affect our core deliberative agency do not understand, because general moral choices made us but understand.