To Humanize or Not to Humanize: Users’ Perception on Interactions with a Mental Health Conversational Agent

Abstract

Given the rapid development of technology accurately simulating human behavior, the need to understand its implications particularly in mental health human-computer interactions becomes acute. Thus, this paper addresses the role of perception in the context of interactions between the cognitive behavioral therapy conversational agent Wysa: Anxiety, therapy chatbot and its users, looking at the way it influences the type of interactions constructed. The study attempts to contribute to the growing research on mental health human-computer interaction by analyzing 150 users’ reviews from Google Play Store using reflexive thematic analysis, through the lenses of the Computers are Social Computers paradigm and the communication scripts perspective. As a result, the study offers the opportunity to reconsider the relevance and usability of theories that emerged several decades ago in the context of technology that simulates human behavior more accurately than at the moment of their appearance. With six developed themes pertaining to the agent’s portrayal and its capabilities, the findings highlight the clash of two opposing tendencies, namely humanizing the agent and reaffirming its robotic nature. These tendencies, shaped by the users’ perception, lead to the construction of distinct interactions, ultimately facilitating a speculative discussion on the issue of authenticity in interactions and the need to reconceptualize it. The findings are in line with previous research but highlight the benefits of viewing perception as an important driving force in the context of the chosen theoretical frameworks.