Towards Interpersonal Assistants: Next-Generation Conversational Agents

We propose novel interpersonal assistants, a next-generation conversational agent which is always-on, unobtrusively serving natural human-to-human conversations. We deepen the motivation and design insights with real practices in language delays and parent–child conflicts. We then present a common platform initiative to effectively support rapid development of interpersonal assistant applications, with a highlight on the key functional element of turn isolations and technical insights on microstructural dynamics.

Conversation would be the most prevailing mode of social interaction. Recently, we have witnessed rapid penetration of intelligent services that listen and respond to human conversations in everyday living space and even on-the-go. Commercial services such as Amazon Alexa, Apple Siri, and Google Assistant are becoming increasingly pervasive. Most of today's services share a common design metaphor—a personal assistant. An intelligent service naturally serves the user by acting as a conversation counterpart.