Don’t role your eyes at me! How anthropomorphic markers influence perceptions of chatbots. A role theory perspective.
(A2020-63662)

AUTHORS

Joseph Ollier, ETH Zürich; Florian von Wangenheim, ETH Zürich

KEYWORDS

text-based chatbot;; self-service technologies; anthropomorphism

ABSTRACT

Chatbots are a new form of service robot, redefining the front line between customer and firm. However, to date, theoretical approaches to anthropomorphism are required to advance the field further. Via the lens of role theory, this paper explores anthropomorphic design considerations in chatbot design, and how this influences downstream service evaluations such as service satisfaction, NPS, co-production and conversational enjoyment. Through four experiments (two-online, two-field) in medical and vehicle rescue settings, we explore how anthropomorphic markers serve as semantic markers illuminating roles in play, and how combinations of these markers serve to reinforce service roles and influence service evaluations. Findings show the relative weight of certain anthropomorphic markers (for example, importance of female avatar for medical setting), or how combinations of markers can be utilized together (formality of communication style and emojis) to foster certain role impressions. Downstream consequences for both firm and customer are explored.