"Hey Siri, Don't Make Me Mad" - Overcoming User Annoyances with Voice Assistants

Author(s):

Strauss, Christina; Harr, Michael; Schütte, Reinhard; Wimmer, Simon

Name of Event:

European Conference on Information Systems (ECIS)

Location:

Paphos, Cyprus

Date:

18.06.2024

Language:

English

Keywords:

Voice Assistants, Large Language Models, User Annoyances, Experiment, Interviews

Presentation:

Download (1.60 MB)

Associated Publication:

"Hey Siri, Don't Make Me Mad" - Overcomming User Annoyances With Voice Assistants

Abstract

This study examines the effects of integrating a technically advanced and more human-like large language model into a voice assistant to assess, how technical advancements mitigate user annoyances. Therefore, a generative pre-trained transformer was integrated into Siri and made available to 23 interview participants. Preliminary results reveal a decrease in user-reported annoyances, showing that the integration not only improves technical accuracy but also enhances the perceived humanness of interactions. However, subsequent interviews indicated that the distinction between the effects of technical advancements and the infusion of humanness emerged as critical, indicating a complex interplay between these factors. It is therefore planned to differentiate between technical and human improvements in the further development of this article. The results contribute to the discourse on optimizing voice assistants by pinpointing the reduction of user annoyances as a pivotal factor in improving user experience, suggesting pathways for future enhancements in voice assistant platforms.