Paper Number
ECIS2026-1226
Paper Type
CRP
Abstract
In recent years, many service providers have used digital humans, which are AI-driven agents with a high degree of humanness. Digital humans are often perceived as more trustworthy. Due to their high technical complexity, errors can occur, raising the question of how errors in digital humans affect benevolence, integrity, and competence. In an online experiment, we find that users attribute significantly more trust to an error-free digital human than to a point-and-click interface (PCI). Contrary to algorithm aversion theory, we find that errors have significantly more negative effects on trust in a digital human than in a PCI. We suggest that trust building is a social process and that the digital human “failed” at it. For theory, we provide evidence that algorithm aversion cannot be observed among users of digital humans. For practice, we suggest that service providers should only implement digital humans if they can ensure their flawless functioning.
Recommended Citation
Mücksch, Janek; Papen, Marie-Christin; Lichtenberg, Sascha; Brendel, Alfred Benedikt; Bellger, Mariam; and Siems, Florian, "Trust Me If You Can: The Role Of Competence, Benevolence, and Integrity When Digital Humans Fail" (2026). ECIS 2026 Proceedings. 4.
https://aisel.aisnet.org/ecis2026/ai_anthro/ai_anthro/4
Trust Me If You Can: The Role Of Competence, Benevolence, and Integrity When Digital Humans Fail
In recent years, many service providers have used digital humans, which are AI-driven agents with a high degree of humanness. Digital humans are often perceived as more trustworthy. Due to their high technical complexity, errors can occur, raising the question of how errors in digital humans affect benevolence, integrity, and competence. In an online experiment, we find that users attribute significantly more trust to an error-free digital human than to a point-and-click interface (PCI). Contrary to algorithm aversion theory, we find that errors have significantly more negative effects on trust in a digital human than in a PCI. We suggest that trust building is a social process and that the digital human “failed” at it. For theory, we provide evidence that algorithm aversion cannot be observed among users of digital humans. For practice, we suggest that service providers should only implement digital humans if they can ensure their flawless functioning.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.