Understanding Circumstances for Desirable Proactive Behaviour of Voice Assistants: The Proactivity Dilemma

Abstract

The next major evolutionary stage for voice assistants will be their capability to initiate interactions by themselves. However, to design proactive interactions, it is crucial to understand whether and when this behaviour is considered useful and how desirable it is perceived for different social contexts or ongoing activities. To investigate people's perspectives on proactivity and appropriate circumstances for it, we designed a set of storyboards depicting a variety of proactive actions in everyday situations and social settings and presented them to 15 participants in interactive interviews. Our findings suggest that, although many participants see benefits in agent proactivity, such as for urgent or critical issues, there are concerns about interference with social activities in multi-party settings, potential loss of agency, and intrusiveness. We discuss our implications for designing voice assistants with desirable proactive features.