Keep it Short: A Comparison of Voice Assistants' Response Behavior

要旨

Voice assistants (VAs) are present in homes, smartphones, and cars. They allow users to perform tasks without graphical or tactile user interfaces, as they are designed for natural language interaction. However, we found that currently, VAs are emulating human behavior by responding in complete sentences, limiting the design options, and preventing VAs from meeting their full potential as a utilitarian tool. We implemented a VA that handles requests in three response styles: two differing short keyword-based response styles and a full-sentence baseline. In a user study, 72 participants interacted with our VA by issuing eight requests. Results show that the short responses were perceived similarly useful and likable while being perceived as more efficient, especially for commands, and sometimes better to comprehend than the baseline. To achieve widespread adoption, we argue that VAs should be customizable and adapt to users instead of always responding in full sentences.

著者
Gabriel Haas
Ulm University, Ulm, BW, Germany
Michael Rietzler
Institute of Mediainformatics, Ulm, Germany
Matt Jones
Swansea University, Swansea, United Kingdom
Enrico Rukzio
University of Ulm, Ulm, Germany
論文URL

https://dl.acm.org/doi/abs/10.1145/3491102.3517684

動画

会議: CHI 2022

The ACM CHI Conference on Human Factors in Computing Systems (https://chi2022.acm.org/)

セッション: Interaction Schemes and Patterns II

293
5 件の発表
2022-05-03 23:15:00
2022-05-04 00:30:00