聊天機器人對話風格與定製化對使用者感受之影響:以薪資談判為例

Abstract

隨著大型語言模型驅動的聊天機器人技術進步，對話的流暢性和靈活性顯著提升，使其在指導角色上展現出巨大潛力，然而，如何建立使用者與聊天機器人之間的信任以及提升使用者感受仍是重要的議題。本研究探討了聊天機器人的對話風格、定製化外觀和命名對使用者信任與感受的影響，特別是在指導薪資談判的情境下。研究結果顯示，非正式對話風格的聊天機器人能增強使用者的信任、社會臨場感、溫暖、享受和使用意圖，並且女性受測者對於非正式對話風格的效果更為顯著。相較之下，在定製化聊天機器人的外觀和命名方面，未發現信任和其他感受上有顯著差異。儘管如此，我們發現頭像的性別和衣著特徵會影響男性使用者的信任與感受。本研究證實了在指導薪資談判的情境下，聊天機器人的對話風格以及頭像特徵對於使用者感受的影響，並結合質化分析結果，提供了對於對話風格和定製化聊天機器人設計的改進建議。這些發現為未來設計和應用聊天機器人提供參考，幫助設計者創造更符合使用者需求和提升信任感的聊天機器人。

With the advancement of chatbot technology driven by large language models, the fluency and flexibility of conversations have significantly improved, showcasing great potential in advisory roles. However, building trust between users and chatbots and enhancing user perceptions remain critical issues. This study explores the impact of chatbot conversation styles, personalized appearances, and naming on user trust and perceptions, particularly in the context of guiding salary negotiations.

The results indicate that informal conversation styles in chatbots can enhance user trust, social presence, warmth, enjoyment, and intention to use, with female participants showing a more pronounced effect. In contrast, we found no significant differences in trust and other perceptions regarding the personalized appearance and naming of chatbots. Nevertheless, we discovered that the gender and attire characteristics of avatars influence male users' trust and perceptions.

This study confirms the impact of conversation styles and avatar characteristics on user perceptions in the context of guiding salary negotiations. Combining qualitative analysis results, we provide recommendations for improving conversation styles and personalizing chatbot designs. These findings offer valuable insights for the future design and application of chatbots, helping designers create systems that better meet user needs and enhance trust.