Toward socially competent AI: Designing multi-user interaction with embodied intelligent agents to support politeness and fairness (SCAI)

Social competence of intelligent agents is currently a key issue in Human-Robot Interaction (HRI) and Human-Computer Interaction (HCI) research. However, social rules for agents´ behavior in multi-user interaction contexts have been so far largely unexplored.

This is a significant limitation, since multi-user interaction with intelligent agents is becoming increasingly common and presents new and substantial challenges.

The SCAI project addresses this limitation by conducring a design-based study of social acceptability of intelligent agents´ behavior in multi-user interaction contexts.

The projects aims to:develop a conceptual framework for social competence of intelligent agents by introducing “politeness” (showing consideration and respect), and “fairness” (acting to treat everyone equally) as key aspects of socially competent behavior, complementing “social awareness”,identify, through User Enactments and empirical user studies, the types of agents’ behavior, which are perceived as “polite” and “fair” in different multi-user contexts,explore, through prototyping and evaluation, interaction design solutions for "polite" and "fair" behavior of intelligent agents.The project employs state-of-the-art concepts and methods in HCI and HRI, and combines an analysis of real-life multi-user interaction with existing intelligent agents (such as social robots and smart speakers) with a proactive exploration of potential future interaction scenarios.