Options
探索AI虛擬角色個性化設計對不同性別學習者情緒之影響
Other Title
Exploring the Emotional Impact of Personalized AI Virtual Character Design on Learners of Different Genders
Date Issued
2025-03-05
Advisor
王雨涵
Abstract
聊天機器人在教育中的應用日益普及,其個性設計被證明能影響使用者的情感反應,而學習中的情緒則影響了認知過程。過去的研究多集中於聊天機器人的資訊傳遞與輔助功能,較少探討其對學習情緒的影響。本研究旨在填補這一空白,探討聊天機器人(以下稱AI虛擬助教)在課堂中的應用,並分析不同個性設定對不同性別學生學習情緒的影響。
本研究設計了兩款AI虛擬助教,分別具有高親和性和低親和性,並根據過去研究設定語氣和內容特徵,以區分其表現。在前期研究中,共邀請6名學習者以及1名外語教育專家,在體驗AI虛擬助教後進行訪談,並根據回饋內容針對虛擬助教對話設定進行調整。本研究採用實驗設計,共招募24位受試者(14位男性和10位女性),在以英語作為主要語言的設計專業課堂中,分別與兩款AI虛擬助教互動一週,並在隨後一週填寫學習成就情緒問卷以收集情緒資料。此外,另招募六位受試者(三位男性和三位女性)進行質性訪談,以了解受試者對AI虛擬助教之使用情緒體驗。問卷結果分析包含描述性統計以及T檢定分析,以確認不同個性設定對不同性別學習者情緒的影響顯著性。質性訪談則以主題分析法進行分析,以深入了解學生對於聊天機器人的不同個性使用體驗和意見,並了解其對使用AI虛擬角色於學習過程中的看法。
研究結果顯示,高親和性虛擬助教在提升正向情緒(如享受、希望),並降低負向情緒(如焦慮、無聊)方面,其效果優於低親和性助教。女性學習者整體表現出較高的正向情緒和較低的負向情緒,而男性在使用兩款助教時的情緒則相較於女性有更加明顯的差異,高親和性助教顯著降低其無聊感,並提升學習的享受與希望。這顯示設計虛擬助教時,應提高親和性設計,特別是針對男性學習者,以改善情緒並提升學習效果。本研究顯示了AI虛擬助教的個性設計在提升學習者情緒的潛力,特別是針對不同性別的需求差異。這為未來虛擬助教的設計提供了方向,促進更具包容性的數位學習環境,進而推動教育與科技的融合。
本研究設計了兩款AI虛擬助教,分別具有高親和性和低親和性,並根據過去研究設定語氣和內容特徵,以區分其表現。在前期研究中,共邀請6名學習者以及1名外語教育專家,在體驗AI虛擬助教後進行訪談,並根據回饋內容針對虛擬助教對話設定進行調整。本研究採用實驗設計,共招募24位受試者(14位男性和10位女性),在以英語作為主要語言的設計專業課堂中,分別與兩款AI虛擬助教互動一週,並在隨後一週填寫學習成就情緒問卷以收集情緒資料。此外,另招募六位受試者(三位男性和三位女性)進行質性訪談,以了解受試者對AI虛擬助教之使用情緒體驗。問卷結果分析包含描述性統計以及T檢定分析,以確認不同個性設定對不同性別學習者情緒的影響顯著性。質性訪談則以主題分析法進行分析,以深入了解學生對於聊天機器人的不同個性使用體驗和意見,並了解其對使用AI虛擬角色於學習過程中的看法。
研究結果顯示,高親和性虛擬助教在提升正向情緒(如享受、希望),並降低負向情緒(如焦慮、無聊)方面,其效果優於低親和性助教。女性學習者整體表現出較高的正向情緒和較低的負向情緒,而男性在使用兩款助教時的情緒則相較於女性有更加明顯的差異,高親和性助教顯著降低其無聊感,並提升學習的享受與希望。這顯示設計虛擬助教時,應提高親和性設計,特別是針對男性學習者,以改善情緒並提升學習效果。本研究顯示了AI虛擬助教的個性設計在提升學習者情緒的潛力,特別是針對不同性別的需求差異。這為未來虛擬助教的設計提供了方向,促進更具包容性的數位學習環境,進而推動教育與科技的融合。
Chatbots are increasingly used in education, with personality design impacting emotions and learning. However, most studies focus on chatbots' support roles, neglecting their emotional impact. This study fills that gap by exploring the use of chatbots (AI virtual assistants) in the classroom and examining how their personality designs affect the emotions of male and female students.
Two AI virtual TAs were designed with high and low affinity, distinguished by tone and content based on prior research. In a preliminary study, six learners and one language education expert tested the virtual TAs and gave feedback to refine their dialogues. In the main experiment, 24 participants (14 males and 10 females) interacted with the AI virtual TAs in an English-language design course for one week, followed by an academic emotion questionnaire. Six participants (three males and three females) also participated in interviews to provide further insight into their emotional experiences. Questionnaire results were analyzed with descriptive statistics and t-tests to assess the impact of personality on emotions across genders, while interviews were analyzed thematically.
Results show that female learners, regardless of the AI virtual TA’s personality, reported more positive and fewer negative emotions than male learners. Males displayed more notable emotional differences between the two TAs than females. This study suggests that AI virtual assistants should be designed with higher friendliness, especially for male learners, to enhance emotional engagement and improve learning outcomes. This provides direction for future virtual assistant design, fostering an inclusive digital learning environment and advancing the integration of education and technology.
Two AI virtual TAs were designed with high and low affinity, distinguished by tone and content based on prior research. In a preliminary study, six learners and one language education expert tested the virtual TAs and gave feedback to refine their dialogues. In the main experiment, 24 participants (14 males and 10 females) interacted with the AI virtual TAs in an English-language design course for one week, followed by an academic emotion questionnaire. Six participants (three males and three females) also participated in interviews to provide further insight into their emotional experiences. Questionnaire results were analyzed with descriptive statistics and t-tests to assess the impact of personality on emotions across genders, while interviews were analyzed thematically.
Results show that female learners, regardless of the AI virtual TA’s personality, reported more positive and fewer negative emotions than male learners. Males displayed more notable emotional differences between the two TAs than females. This study suggests that AI virtual assistants should be designed with higher friendliness, especially for male learners, to enhance emotional engagement and improve learning outcomes. This provides direction for future virtual assistant design, fostering an inclusive digital learning environment and advancing the integration of education and technology.
Subjects
聊天機器人
人工智慧
學習情緒
個性化虛擬助教
Chatbot
Artificial Intelligence
Learning Emotions
Personalized Virtual Assistant
Type
master thesis