Phenotropic Interaction in Virtual Assistants

Virtual assistants serve as popular conversation-based interfaces, enabling hands-free interactions and pervasive control of home automation systems. To enhance customization, virtual assistants often offer the ability to program customized behaviors. These skills are based on IF This Then That (IFTTT) rules, which dictate the execution of tasks based on observed events or commands. However, virtual assistants often struggle with accuracy in understanding and retrieving information in these personalized rule scenarios where specific wording must be replicated exactly. This chapter explores techniques to improve virtual assistant interactions by implementing phenotropic interaction design principles. By applying techniques from previous chapters, this chapter demonstrates an automated approach to extending IFTTT rules and applying them in similar situations. This technique aims to create a more natural interaction between users and virtual assistants by enabling reasoning and understanding of queries beyond exact matching. The chapter introduces the Flexible Virtual Assistant prototype, which embodies these enhancements and demonstrates the potential of phenotropic human–computer interaction based on a conversational interface. This prototype is used as the basis for a user experiment that highlights the benefits of phenotropic design principles in enhancing naturalness, robustness, and user satisfaction with an interactive system.