Skip to Content
Behavioral SciencesBehavioral Sciences
  • Article
  • Open Access

10 February 2026

15 Pages

AI Communication Tone and Consumer Judgment: The Role of Servant Perception in Behavioral Intentions

Business School, Hankuk University of Foreign Studies, Seoul 02450, Republic of Korea

Abstract

Artificial intelligence is increasingly embedded in service interactions, requiring users to form rapid social judgments about AI communicators based on limited linguistic and contextual cues. This research examines how AI communication tone shapes behavioral intentions through social cognitive processes of role construal and agency attribution. Drawing on politeness theory, formality research, and social cognition perspectives, two scenario-based experiments test whether formal versus casual tone influences responses via attitudes toward the tone and the AI, and how these effects depend on perceptions of AI as a servant-like social actor. Study 1 shows that tone effects are moderated by servant perception and that economic framing, specifically paid versus free access, functions as an antecedent of hierarchical role construal. Study 2 replicates these effects and demonstrates that interaction structure, one-way versus two-way communication, similarly shapes servant perception by signaling differential autonomy. Across both studies, formal tone is more effective when AI is construed as subordinate, whereas casual tone is less effective under hierarchical role frames. By identifying servant perception as a central social cognitive mechanism, this research advances understanding of human judgment and decision making in technology-mediated interactions and offers implications for AI communication design aligned with role expectations. Because both studies rely on U.S. consumers, the findings should be interpreted within cultural contexts characterized by relatively low power distance, where role expectations and hierarchy norms may differ from other cultural settings.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.