The Agreeableness of a Virtual Agent: Effects of Reciprocity and Timing of Help

EasyChair Preprint 4445

Abstract

The virtual agent’s agreeableness has been highlighted with the increasing importance of social interaction with intelligent agents. As a virtual assistant, the agents are perceived as agreeable by users with their support of help and cooperation. Considering the context-dependent nature of human-agent interaction, it is important to examine the situational variables in defining an agent’s behavior as a predictor of agreeableness. In this regard, our study conducted 2 (help type: reciprocated vs. unreciprocated) x 2 (timing of help: wanted vs. unwanted) between-subjects design to analyze the effect of agent’s reciprocity of help as well as its timing of whether the recipients want help or not. Our results show that an intelligent agent giving unreciprocated help was perceived as much agreeable and socially attractive than an agent showing reciprocated help. Also, the recipient’s unwanted help damaged users’ perception of the agent’s agreeableness and social attraction. These findings have both theoretical and practical implications.

Keyphrases: Agent personality, CASA, Reciprocity norm, social attraction, virtual agent

@booklet{EasyChair:4445,
  author    = {Jieun Kim and Young June Sah and Hayeon Song},
  title     = {The Agreeableness of a Virtual Agent: Effects of Reciprocity and Timing of Help},
  howpublished = {EasyChair Preprint 4445},
  year      = {EasyChair, 2020}}