EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

When friendly AI backfires: Relationship norm misfit in chatbot service recovery

Duy Khanh Phan, Bao Quoc Truong-Dinh

Computers in Human Behavior Reports · 2026-12 · 24 · 101409

DOI: 10.1016/j.chbr.2026.101409 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computers in Human Behavior Reports, 2026-12. Verfasst von Duy Khanh Phan und 1 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study identifies a context-specific boundary condition within established frameworks governing chatbot service recovery: the effectiveness of warm, relational recovery cues depends on alignment with the relationship norms operative in the exchange context. Drawing on Relationship Norms Theory and Justice Theory, we develop and test a norm-contingent inference account in which recovery effectiveness is governed by alignment between chatbot framing and the dominant relationship logic of the exchange. Across four controlled experiments progressing from an existence proof through mechanism isolation and boundary diagnosis to a high-fidelity behavioural replication we demonstrate a robust friendly-recovery backfire: communal chatbot framing reduces forgiveness, repurchase intention, and behavioural acceptance when exchange norms are salient. Perceived fairness emerges as the primary operative mechanism, mediating these effects and excluding anthropomorphism, responsibility attribution, and perceived control as alternative pathways. Relationship norm salience functions as a moderating boundary condition that reverses the evaluative meaning of identical recovery cues, producing sign-changing indirect effects across contexts. The contribution is positioned as a boundary-condition specification and mechanism-level extension within Relationship Norms Theory and Justice Theory applied to AI-mediated chatbot service recovery: specifying when norm logic renders warm recovery cues inappropriate, with perceived fairness as the operative mechanism and norm salience as the contextual moderator. The findings extend and integrate established theories to explain sign-reversing effects in AI-mediated digital commerce, offering a falsifiable, design-relevant account of chatbot recovery governance within the scope of four experiments using exchange-oriented e-commerce failure scenarios.

Zitieren

APA

Phan, D. K., & Truong-Dinh, B. Q. (2026). When friendly AI backfires: Relationship norm misfit in chatbot service recovery. Computers in Human Behavior Reports, 24, 101409. https://doi.org/10.1016/j.chbr.2026.101409

Vancouver

Phan DK, Truong-Dinh BQ. When friendly AI backfires: Relationship norm misfit in chatbot service recovery. Computers in Human Behavior Reports. 2026;24:101409. doi:10.1016/j.chbr.2026.101409

BibTeX
@article{Phan2026,
  author = {Phan, Duy Khanh and Truong-Dinh, Bao Quoc},
  title = {When friendly AI backfires: Relationship norm misfit in chatbot service recovery},
  journal = {Computers in Human Behavior Reports},
  year = {2026},
  volume = {24},
  pages = {101409},
  doi = {10.1016/j.chbr.2026.101409}
}
Kostenlos starten →