City Research Online

Abstract

Current thrusts in explainable AI (XAI) have focused on using interpretability or explanatory debugging as frameworks for developing explanations. We argue that for some systems a different paradigm – persuasive engagement – needs to be adopted, in order to affect trust and user satisfaction. In this paper, we will briefly provide an overview of the current approaches to explain smart systems and their scope of application. We then introduce the theoretical basis for persuasive engagement, and show through a use case how explanations might be generated. We then discuss future work that might shed more light on how to best explain different kinds of smart systems.

Publication Type:	Article
Additional Information:	© 2019 for the individual papers by the papers' authors. Copying permitted for private and academic purposes
Publisher Keywords:	Explanations; explanatory debugging; intelligibility; smart systems, intelligent user interfaces; argumentation
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Departments:	School of Science & Technology > Computer Science

Export

Downloads