Effects of Embodiment on Generic and Content-Specific Intelligent Virtual Agents as Exhibition Guides

Abstract

Intelligent Virtual Agents (IVAs) received enormous attention in recent years due to significant improvements in voice communication technologies and the convergence of different research fields such as Machine Learning, Internet of Things, and Virtual Reality (VR). Interactive conversational IVAs can appear in different forms such as voice-only or with embodied audio-visual representations showing, for example, human-like contextually related or generic three-dimensional bodies. In this paper, we analyzed the benefits of different forms of virtual agents in the context of a VR exhibition space. Our results suggest positive evidence showing large benefits of both embodied and thematically related audio-visual representations of IVAs. We discuss implications and suggestions for content developers to design believable virtual agents in the context of such installations.