Are virtual influencers falling into the uncanny valley? An empirical study perspective

Abstract

Virtual influencers (VIs) are computer-generated avatars that can often be indistinguishable from a human. Many of them are controlled by humans but when controlled by Artificial Intelligence (AI) they hold enormous potential due to its low cost and capability to scale. As VI are an emerging technology, it is important to explore how VIs controlled by AI vs human-controlled are perceived by users, and how their anthropomorphic appearance may enhance users’ favorability towards the entity. Based on human-computer interaction research and on the Uncanny Valley Theory, this study applies a social relationship perspective to the design and appliance of VIs on the Instagram environment by running an online experiment using a 2x2 factorial design. The study was conducted on an international sample of 174 participants. Results shows that, as for anthropomorphic appearance, human-like VIs are perceived as more favorable than cartoonlike ones, but when considering if VIs are controlled by AI, users do not show any significant difference on their interpersonal feelings. The findings give a broader understanding of VIs adoptions since the study represents one of the first attempt to provide insights from an experiment including VIs appearance (human-like vs cartoon-like) and agency (human controlled vs AI controlled).

Date of Award	11 Jul 2022
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Susana Silva (Supervisor) & Roberta De Cicco (Co-Supervisor)