Modeling Consequences of Brand Authenticity in Anthropomorphized AI-Assistants: A Human-Robot Interaction Perspective

Abstract

The emergence of anthropomorphized AI Assistants can be linked to the advanced convergence of machine learning and natural language processing algorithms that could mimic human brains. Conversational-AI has led users to expect a sense of authenticity in their anthropomorphized assistants, more so, in a social context; which creates newer avenues for brands to better connect with their consumers. The present study aimed to develop a consequential model of AI-authenticity while drawing inferences from a series of human-robot interaction based theories, viz. “Computers as Social Actors” (CASA); “Media Equation” (ME), “Stereotype Content Model” (SCM) and “Socio-Cognitive Computational Trust” (SCCT) theory. Partial-Least-Square based Structural-Equation-Modeling was performed to examine the hypothesized framework; while, bootstrapping technique was utilized to better assess the effect of mediation analysis. The predictive relevance of the developed model was evaluated based on cross-validated redundancy approach. The findings designated ‘Emotional Attachment’, ‘Customer Engagement’ and ‘Cognitive Trust’ as major consequences of brand authenticity; while ‘warmth’was accounted as a positive, but weak mediator in authenticity-cognitive trust relationship, due to probable effects of uncanny valley phenomenon. ‘Cognitive Trust’remained a significant predictor of ‘continuous usage intentions’and ‘word-of[1]mouth’ behaviour. The proposed AI-authenticity framework could aid underpinning effective customer retention and extension strategies.