ABSTRACT:
A digital twin is an artificial intelligence (AI)- or human-controlled avatar that replicates a specific person’s identity. While organizations have used digital twins of employees (e.g. digital doctors answering patients’ questions), a more common use is celebrity-as-a-service, where digital twins of celebrities interact with customers. Unlike traditional celebrity endorsements in marketing videos, these digital-twin agents interact directly with consumers. Although digital twins do not enhance the underlying AI or human agents’ capabilities, their appearance as a specific person can influence users’ perception. Across two experiments, we found that when a digital twin resembled a celebrity, regardless of whether it was AI- or human-controlled, it was perceived to be more capable, benevolent, and trustworthy, leading to greater user engagement. Even when errors occurred, the celebrity’s likeness reduced negative reactions, suggesting that celebrity effects can buffer against algorithm aversion. Practically, digital celebrity twins offer business value comparable to traditional celebrity endorsements, influencing user trust and service adoption despite identical underlying performance.
Key words and phrases: Digital human, celebrity endorsement, algorithm aversion, trustworthiness, digital twins, celebrity-as-a-service, online agents