Scholarlink

<p>This study explores user politeness in task-oriented information-seeking interactions with generative AI systems, focussing on how users deploy politeness strategies, how these interactional behaviours reflect their perceptions of AI and whether the politeness patterns they exhibit are influenced by the demographic factors age and gender. Using a subset of Frummet et al.’s (ACM Trans Inf Syst 42(5):1–29, 2024) <i>Cooking with Conversation</i> dataset, we annotated 30 user-agent conversations for politeness-relevant speech acts and conducted a cluster analysis. We derived four distinct politeness clusters: <i>Hyperpolite</i>, <i>Polite and engaged</i>, <i>Engagement-seeking</i> and <i>Hyperefficient</i>, reflecting a spectrum of interactional styles ranging from highly respectful, appreciative and engaged to task-oriented and efficient. The clusters suggest varying user perceptions of the agent, from human-like conversational partner to a more machine-like functional tool. However, no significant effects of age or gender on politeness patterns were found. This research establishes a basis for future interdisciplinary studies on how variation in user politeness—and the degree to which it is mirrored by the system—influence user satisfaction, inclusivity and information transfer.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

From Hyperpolite to Hyperefficient: A Continuum of User Politeness in Human–GenAI Interactions

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月13日

查看全文
  • Christine Elsweiler,
  • Anna Ziegner,
  • David Elsweiler

摘要

This study explores user politeness in task-oriented information-seeking interactions with generative AI systems, focussing on how users deploy politeness strategies, how these interactional behaviours reflect their perceptions of AI and whether the politeness patterns they exhibit are influenced by the demographic factors age and gender. Using a subset of Frummet et al.’s (ACM Trans Inf Syst 42(5):1–29, 2024) Cooking with Conversation dataset, we annotated 30 user-agent conversations for politeness-relevant speech acts and conducted a cluster analysis. We derived four distinct politeness clusters: Hyperpolite, Polite and engaged, Engagement-seeking and Hyperefficient, reflecting a spectrum of interactional styles ranging from highly respectful, appreciative and engaged to task-oriented and efficient. The clusters suggest varying user perceptions of the agent, from human-like conversational partner to a more machine-like functional tool. However, no significant effects of age or gender on politeness patterns were found. This research establishes a basis for future interdisciplinary studies on how variation in user politeness—and the degree to which it is mirrored by the system—influence user satisfaction, inclusivity and information transfer.