Making emotional transparency transparent

Abstract From LLM-based artificial friends to social robots, an increasing number of artificial agents are explicitly designed to interact with us in an emotionally salient way. The notion of emotional transparency has recently been proposed to describe the mechanisms behind users’ emotional involvement with such systems. However, the details and theoretical boundaries of emotional transparency have been left largely unspecified. This article provides an in-depth analysis of emotional transparency. First, emotional transparency is characterized as a form of transparency in use and situated in the current theoretical landscape of varieties of transparency in the philosophy of mind and technology. Then, the drivers of emotional transparency are investigated, focusing on the features of artificial systems, users, and context of use. Finally, the epistemic state of users interacting with emotionally transparent artificial agents is analyzed, focusing on the apparent paradox by which users behave as if such agents had emotions even if they are perfectly aware that this is not the case.