Understandable and trustworthy explainable robots: A sensemaking perspective
Guglielmo Papagni, Sabine T. Koeszegi
Abstract
Abstract This article discusses the fundamental requirements for making explainable robots trustworthy and comprehensible for non-expert users. To this extent, we identify three main issues to solve: the approximate nature of explanations, their dependence on the interaction context and the intrinsic limitations of human understanding. The article proposes an organic solution for the design of explainable robots rooted in a sensemaking perspective. The establishment of contextual interaction boundaries, combined with the adoption of plausibility as the main criterion for the evaluation of explanations and of interactive and multi-modal explanations, forms the core of this proposal.
Topics & Concepts
SensemakingPerspective (graphical)TrustworthinessRobotContext (archaeology)Computer scienceHuman–computer interactionHuman–robot interactionKnowledge managementArtificial intelligenceEpistemologyData scienceCognitive sciencePsychologyInternet privacyPhilosophyBiologyPaleontologyExplainable Artificial Intelligence (XAI)Ethics and Social Impacts of AIReinforcement Learning in Robotics