From Apology to Compensation: A Multi-level Taxonomy of Trust Reparation for Highly Automated Virtual Assistants

摘要

This paper presents a multi-level taxonomy of reparation levels specifically adapted to virtual assistants in the context of Human-Human-Interaction (HHI) with a specific focus on maintaining trust in the system. This taxonomy ranges from current models of apology to the newly integrated compensation area via a range of case studies specifically developed to address the rising concerns of unsupervised interactions in the context of Virtual Assistants (VA). Based on preliminary research, the author recommends the integration of reparation strategies as a fundamental variable in the ongoing development of VAs, as this element inserts a sense of balance in terms of vulnerability between users and developers to enhance trust in the interactive process. Present and future work is being dedicated to further understand how different contexts may affect integrity in highly automated virtual assistants.