Customer-service agents rendered as digital twins of a celebrity were rated more capable, more benevolent and more trustworthy than identical agents without the likeness, according to a study in the Journal of Management Information Systems (vol. 43, no. 2, 2026) covered by Phys.org on October 28, 2026. The effect held whether the twin was controlled by an AI or by a human, and it raised users’ engagement.
Why it matters: brands are starting to license celebrity faces for AI agents. The study puts a measured effect on what that likeness buys.
Key Facts
- Paper: “Celebrity-as-a-Service”, Journal of Management Information Systems 43(2), 2026, pp. 335–367
- Finding: celebrity-likeness twins rated more capable, benevolent and trustworthy, with higher engagement
- Errors: likeness reduced negative reactions after mistakes
- Holds for both AI- and human-controlled twins; underlying performance identical
What did the researchers find about errors?
The celebrity likeness reduced negative reactions after the agent made mistakes. The authors read this as evidence that celebrity effects can offset “algorithm aversion”, the tendency to lose trust in automated systems faster than in people after an error. Performance was identical across conditions; only appearance changed.
Who wrote it?
Lingyao (Ivy) Yuan, Antino Kim, Mike Seymour and Alan R. Dennis, in a paper titled “Celebrity-as-a-Service: Trust and Willingness to Use Digital Twins for Customer Service” (pp. 335–367). They conclude that digital celebrity twins can deliver business value similar to traditional celebrity endorsements.
Where does regulation stand?
Rules on synthetic likeness are tightening. The EU AI Act’s transparency duties cover AI-generated faces (our explainer), and Beijing has begun issuing ID cards for digital humans. See also celebrity AI avatars.
Frequently Asked
Do celebrity AI avatars make customers trust a service more?
A 2026 study in the Journal of Management Information Systems found celebrity-likeness digital twins were rated more capable, benevolent and trustworthy than identical agents, and increased engagement.
Do celebrity avatars reduce the impact of AI mistakes?
The study found the celebrity likeness reduced negative reactions after the agent made errors, which the authors link to offsetting algorithm aversion.