How the Anthropormorphization of Virtual Assistants Influences User’s Trust
Tim Crone, Mahmood Shafeie Zargar · Academy of Management Proceedings · 2018
Companies are increasingly using Virtual Assistants to communicate with their customers as the AI-driven chatbot technologies gain in maturity. Today’s Virtual Assistants can hold human-like conversations and even start to emulate the looks of human agents. However, research has shown little interest in what this means for the human-automation interaction. Do we trust these Virtual Assistants? And are we willing to rely on their output? In order to address this gap, the current study seeks to gain a better understanding of the drivers of trust in automation, automation reliance, and thereby usage intentions. An experiment is conducted with two manipulations: The behavior of the Virtual Assistant and its looks (avatar) differed among conditions, while its functional performance was held steady across the conditions. The results reveal increased trust in and reliance on Virtual Assistants when they show human-like behaviors, despite the lack of improvement in objective performance. More interestingly, the increase in trust is significant in both affective and cognitive dimensions. The fuzzy feeling of dealing with well-behaved machines seems to be both emotionally pleasing and intellectually convincing.