Usability Evaluation of Artificial Intelligence-Based Voice Assistants: The Case of Amazon Alexa

Abstract

Currently, the use of voice-assistants has been on the rise, but a user-centric usability evaluation of these devices is a must for ensuring their success. System Usability Scale (SUS) is one such popular usability instrument in a Graphical User Interface (GUI) scenario. However, there are certain fundamental differences between GUI and voice-based systems, which makes it uncertain regarding the suitability of SUS in a voice scenario. The present work has a twofold objective: to check the suitability of SUS for usability evaluation of voice-assistants and developing a subjective scale in line with SUS that considers the unique aspects of voice-based communication. We call this scale as the Voice Usability Scale (VUS). For fulfilling the objectives, a subjective test is conducted with 62 participants. An Exploratory Factor Analysis suggests that SUS has a number of drawbacks for measuring the voice usability. Moreover, in case of VUS, the most optimal factor structure identifies three main components: usability, affective, and recognizability and visibility. The current findings should provide an initial starting point to form a useful theoretical and practical basis for subjective usability assessment of voice-based systems.