導入聊天機器人輔助「諮商技巧」課程教學之效果研究

The Effectiveness Study of a Chatbot-Assisted Counseling-Skills Course

本研究旨在探討「導入聊天機器人」輔助諮商技巧課程教學的教學效果，了解諮商技巧課程對學習者諮商技巧學習的影響。本研究採取前實驗研究法之單組前後測設計，資料分析採混合方法進行研究。研究參與者為諮商學系大三學生，研究者在諮商技巧課程前後以「個別諮商能力評量表」與「傑佛遜同理心量表」為評量工具，輔以教室觀察、同理心反思札記了解諮商技巧的訓練效果，並於期末考後邀請其同意進行焦點團體訪談作為質性分析。研究發現經過一學期導入聊天機器人輔助「諮商技巧」課程教學，學習者個別諮商能力除了人格傾向之外，都有長足的進步，包括專業行為、深層互動、訊息互動、基本溝通、概念形成技能、個人化技能、理論基礎等均達顯著差異。其中同理心在不同量表結果表現不一致的現象值得進一步探討。質性分析發現聊天機器人做為模擬個案能降低學習者的練習成本並提高自主學習動機，聊天機器人系統引導學習者逐一認識與練習不同的諮商技巧，減輕練習諮商的焦慮不安，並更熟稔的運用諮商技巧於課堂同儕模擬諮商中，結果顯示導入聊天機器人輔助「諮商技巧」課程教學有其特殊貢獻、效果與可行性。最後本研究針對研究結果提出反思與建議，供未來運用人工智慧輔助諮商教學之依據。

Counselor training requires not only the acquirement of knowledge but also practices of professional skills and techniques. A counseling-skills course usually provides both the knowledge and practice of counseling skills. Due to ethical concerns, the students in the class often use their peers as their practice clients. However, there are three limitations when using peers as clients. First, its availability can be limited because it needs schedule coordination with a peer to set up a practice session. Secondly, it can create pressure to watch and be watched while conducting peer practices. The pressure generates anxiety and can reduce motivation to learn. Lastly, it can be stressful for a novice student to handle the spontaneity of a live counseling session. To address these three limitations, the authors take the advantage of the latest development of artificial intelligence (AI) to incorporate a chatbot into a counseling-skills course. The chatbot techniques have been used in medical, educational, and counseling training. This study used the chatbot as a virtual client for the students to practice their counseling skills. The chatbot system arranges each segment that focuses on an individual skill. It contains instructions to the addressing counseling skill so that the students can reviews before and after their chatbot practice. The chatbot system includes 14 segments which include counseling skills such as debriefing, emotional reflecting, empathizing, structuring, summarizing, immediacy, self-disclosing, and confronting/challenging. After the practice, the chatbot system provided instant feedbacks to the students. The students practiced a chatbot segment each week. It is a self-learning experience since the students implement it from their own electronical devices and in their own free time. The chatbot system allows great availability because the students can get online and practice with the chatbot any time they want. The students are interacting with the virtual chatbot instead of their peer, so the chatbot exercise can reduce the practice-induced stress because. Furthermore, the chatbot arranges practice segments on a counseling skill each session. This allows the students to focus on learning an individual skill instead of dealing the complication and variation of the spontaneous reactions from the peer practice.
This study is to investigate the effectiveness of the chatbot-assisted counseling-skills course. A pre-experimental design was employed. It is with a single-group pretest-posttest methodology and utilizes a mixed-methods approach for comprehensive data analysis. There are 35 third-year undergraduate counseling students participating in this study. This sample has 9 male participants. Both qualitative and quantitative data were collected to examine the effectiveness of the chatbot-assisted course. The quantitative data was from the participants’completions of the Individual Counseling Competence Assessment Form and the Jefferson Scale of Empathy before and after the course. The Individual Counseling Competence Assessment Form has 8 subscales, which are 7 competence subscales (i.e., Professional Demeanors, Deep Exchanges, Information Exchanges, Basic Communication, Conceptualization, Individualization, and Theoretical Comprehension) and a personality subscale which is Professional Personality. The three factors of the Jefferson Scale of Empathy are adopted. They are Perspective Taking, Compassionate Care, and Walking in Patient’s Shoes. The qualitative data was from the focus group discussions at the end of the semester, classroom observations and the students’empathy reflection journals.
The t-tests were performed on the quantitative data. The results show the significant increases in all the 7 counseling competence subscales of the Individual Counseling Competence Assessment Form, but there is no significant change of the personality subscale. For the Jefferson Scale of Empathy, the only significant change is Compassionate Care and it decreased significantly. This may appear to be a counter-effectiveness result, but it might reflect that the novices tend to overestimate their ability to attend to others’emotions and the practices have given them a realistic estimation of their compassionate care for the clients. Overall, the results suggest that the chatbot-assisted counseling-skills course enhances the students’counseling competency and has provided the students with realistic understanding of their own competency, but it has no significant effect on their personality change.
In the qualitative data, the students reported the chatbot practice is one main source of their learning in addition to the class lectures, the peer practices, the transcript writing, and reflection journals. The immediate feedbacks from the chatbot system promoted the students’learning motivation to strengthen their counseling skills. The chatbot allows the students to practice what they had learned from class. It also presents the students with new concepts which stimulate them to search for new counseling knowledge on their own. The chatbot is friendly to the novices because there is no peer pressure and stress. The chatbot program contained abundant client information, provided systematic guidelines, and allowed the students to practice each skill at a time. The students liked the chatbot system because it was an auxiliary learning aid and it is convenient learning avenue for them to practice in their own time. It was efficient because the chatbot system was constantly revised based on the students’feedbacks. However, the students reported the chatbot has limitation of spontaneity. It cannot have facial expressions as vivid as a peer client. Furthermore, the chatbot system cannot respond to the actions that are not programed in it. The students also suggested the increase of synchronicity between the class materials and the chatbot practice content. These results suggest the chatbot contributed the effectiveness of the counseling-skills course through its availability, flexibility, immediacy, and systemic structure. It also reduces stress and promotes self-learning motivation. To advance the chatbot, the system needs to improve its colloquial words and to have human-like aminations. It needs to increase the coordination between the class and the chatbot practices.
Practice is a key component when the students learn counseling skills. However, the peer practice, which is often implemented in a counseling-skills class, can be stressful for the students and requires the schedule coordination with peers. The latest artificial intelligence technology allows the computer to act as a chatbot to interact with people. This study used the chatbot as a virtual client for the students to practice their counseling skills at their own time. The study results provide empirical supports for that the chatbot-assisted learning is a convenient, low-cost and low-stress method to effectively enhance the students’counseling competency and learning motivation. It also reveals the limitations of the chatbot system. These results encourage incorporation of artificial intelligence into the counselor education. The addition of the chatbot system does not only allow the students to practice what they have learned from class, but also contains new materials to promote self-learning motivation. However, this does not recommend the chatbot to completely replace the peer practice since the peer practice provides the students spontaneous human contact experience. The study results suggest modifications for the chatbot system. The chatbot system needs to be more human-like and to incorporate more colloquial words. In addition to the segment practice of an individual skill, the students recommend the system to include complete sessions and continuous clients, so that they can experience the complete process of a session and a counseling case. Though the new materials from the chatbot can promote self-learning, it is suggested to increase the synchronicity between the class and the chatbot system to enhance learning comprehension. Finally, the further researches can focus on different student populations, such as master’s and doctoral counseling students.