Session 12

AI and Social Work Education

Moderator: Dr. Qiuling An (Professor, East China Normal University, China)

O12.1  When Proposals Are "Generated": The Visibility and Invisibility of Service Planning Competence Among Macao MSW Students in the Age of AI

*Weixiao Zhu¹, Yuhong Zhu¹, Cheng Ian Cheong²

¹School of Social Research, Renmin University of China, Beijing, China; ²MindMed Clinic, Macau, China

Abstract

Background and Purpose: The rapid proliferation of generative artificial intelligence (AI) is profoundly reshaping the learning ecosystem in higher education, with social work education also caught in this wave of transformation. As a practice-rooted discipline, social work regards service planning competence as one of its core competencies, requiring students to be able to design operable intervention plans, and internalize professional values and humanistic care into the practice process. However, the convenience of AI tools also harbors a latent risk that warrants vigilance: students may be inclined to directly utilize AI-generated content that appears formally standardized, thereby undermining the acquisition of their core competencies amidst technological dependency. Currently, academic discussions on this issue largely remain at the theoretical and speculative level, lacking empirical analysis based on real student assignment texts, which makes it difficult to reveal the actual impact of AI intervention in the learning process. In light of this, this study takes the final assignments of first-year Master of Social Work (MSW) students at University B in Macao in mental health-related courses as the analytical objects, aiming to address the following core questions: (1) What is the current status of AI technology usage in student assignments? (2) What differences exist in textual characteristics between texts with high and low AI participation? (3) What challenges does the widespread application of AI technology pose to the core values of social work education, and how should educators respond?

Methods: This study adopts a mixed-methods design. The data originates from all final assignments of first-year MSW students at University B in Macao in mental health-related courses during the fall semester of 2025, totaling 21 assignments. The quantitative part employs the CNKI AI detection tool to evaluate the AI detection rates of all texts. The qualitative part utilizes maximum variation sampling to select three assignments with the highest and lowest AI detection rates, respectively, for comparative analysis, focusing on differences between the two types of texts in dimensions such as problem identification, theoretical application, service design and evaluation indicators.

Results: Preliminary analysis reveals that AI has deeply intervened in the assignment writing process of MSW students, with some students showing a strong tendency to rely on AI to generate core content, resulting in significantly higher AI detection scores for their assignments. Assignments with high AI participation excel in structural neatness, normative terminology usage, and richness of literature citations, but they lack observations of Macao's local service system. In contrast, assignments with low AI participation retain stronger personal imprints and situational embeddedness, reflecting a visceral understanding of the complexity of practice, but they also suffer from issues such as inadequate structural neatness and insufficient literature citations. These differences demonstrate that while AI can indeed help students enhance the formal professionalism of their assignments, its role in promoting deep competencies such as independent judgment, critical thinking, and contextualized design of complex social issues remains worthy of scrutiny.

Conclusions and Implications: In response, educators need to adopt a proactive yet prudent response strategy: First, re-examine the evaluation criteria for service planning assignments. Second, guiding students to critically analyze the limitations of AI-generated content and emphasizing students' subjectivity in thinking during the plan design process. Third, strengthen practice-oriented teaching methods, to enhance students' immersive experience in real service scenarios. Finally, establish transparent AI usage norms and ethical guidelines to help students uphold the bottom line of professional values in the technological era.

O12.2  An AI Animated Design Web-Based Single-Session Intervention of Growth Mindsets for Reducing Practicum-Related Anxiety Among Social Work Student: A Randomized Controlled Trial

*Shimin Zhu¹, Yongyi Wang¹

¹The Hong Kong Polytechnic University, Hong Kong, China

Abstract

Background and Purpose: Practicum in social work is intensive, and trainees’ mindsets shape learning experiences. We developed an AI-assisted Design intervention named Web-based Single-session Intervention of Mindsets on Intelligence, Failure, and Emotion (We-SMILE) and have examined its acceptability and efficacy on reducing practicum-related anxiety with a pilot study in 2024. We aimed to examine its effectiveness in a larger-scale randomized controlled trial. We hypothesize that 1) We-SMILE is more effective in reducing anxiety related to practicum among social work a trainees compared to the training as usual(TAU) group (primary outcome) and 2) secondary outcomes, including (1)relieving depression, anxiety, and stress; (2) improving psychological well-being; (3) enhancing learning orientation; and (4) increasing academic self-efficacy and confidence related to practicum compared with the TAU group.

Methods: Participants are recruited from social work programs in eight universities in Hong Kong with a two-arm randomized controlled trial. Participants are randomly allocated to the We-SMILE or training-as-usual (TAU) group. The outcomes will be measured by validated items and scales on anxiety, mindsets, psychological well-being, and the Failure Mindset Scale. Participants are surveyed online at pre- (T0), post-intervention (T1), two-week (T2), and eight-week (T3) follow-ups. Recruitment has started in September 2025 during social work pre-practicum briefing sessions. Data collection were completed by the end of 2025. The intention-to-treat (ITT) analysis principle and linear regression–based maximum likelihood multilevel models will be used for data analysis.

Results: Totally, 270 social work practicum trainees from eight universities participated the trial and were randomly assigned to We-SMILE or TAU group via intervention system. Comparing to the TAU group, the We-SMILE intervention significantly reduced the anxiety about practicum and learning, stress level, and enhanced the confidence, resilient coping and mastery-learning orientation in two-week and eight-week follow-ups. The AI animated design and co-production methods made the intervention engaging and sustained effects of We-SMILE.

Conclusions and Implications: Integrated growth mindset intervention is effective in reducing practicum-related anxiety and enhancing confidence in social work practicum trainee. Web-based single session psycho-education mode is a scalable and user-friendly way to provide intervention for students in need efficiently. Using AI-animation and co-production for development of single-session intervention is a promising way for producing engaging intervention for social work education. We also share the experience in design and implementation of AI-Assisted Design for web-based interventions for future studies.

O12.3  An Applied Research of AI Virtual Client in Practical Skills Training for Social Workers

*Chang Zhuo¹, Liu Jinghong²

¹School of Social Development and Public Policy, Fudan University, Shanghai, China; ²School of Sociology, Nankai University, Tianjin, China

 Abstract

In the field of practical education, social work has been facing structural challenges such as an overemphasis on theory and a neglect of practice, limited training opportunities and high training costs. To address these issues, this study developed an AI virtual client with localized contextual settings based on Qwen which is a large language model in mainland China, aiming to support practical skills training for social workers. The study adopted experimental approaches and focus group interviews with a small-sample design. Twenty-five social work students and novice practitioners were recruited and divided into two groups: Group A, consisting of participants without prior experience interviewing real clients, and Group B, consisting of those with such experience. Both groups engaged in a three-day simulated interview training program. Data were collected through questionnaires measuring perceived realism of the AI virtual client, improvement in practical interview skills and willingness to recommend the AI virtual client to others, as well as through focus group interviews exploring participants’ experiences. Independent-samples t-tests and Mann–Whitney U tests were conducted to examine between-group differences. The results indicated no statistically significant differences between the two groups across the three measured dimensions (p > 0.05), suggesting homogeneity in their evaluations. Both groups reported relatively high mean scores, indicating that participants perceived the AI virtual client as realistic, effective in improving practical interview skills and worthy of recommendation. Qualitative findings further demonstrated that the AI virtual client was particularly effective in strengthening structured interviewing, information gathering and problem-solving skills, while limitations remained in emotional resonance, nonverbal communication, sense of presence and long-term relationship building. Overall, the study concludes that AI virtual client can serve as an effective supplement to traditional practical education and training models of social work, offering a low-risk and highly accessible digital pathway for skills development. These findings provide practical implications for integrating generative AI into social work education and highlight the need for further technological refinement and ethical consideration in future research and implementation.

O12.4  AI and Social Work Education in the Greater Bay Area

*Prof Kam Tong Chan¹

¹Guangdong Open University, China

Abstract

In recent years, the rapid advancement of artificial intelligence (AI) has transformed numerous sectors, yet its integration into social work education and services remains limited. This gap presents a significant opportunity to enhance the effectiveness and reach of social services through innovative technological applications. A collaborative project across the Great Bay Area—involving Guangdong Open University, the Chinese University of Hong Kong, and Macao Polytechnic University—aims to align AI with key areas of social work. These include services for children, the elderly, patient rehabilitation, and community governance. Through action research and partnerships between academics and practitioners, the initiative seeks to develop practical AI-driven tools tailored to specific community needs. In children’s services, the focus is on supporting those with dyslexia through an app designed to improve Chinese character recognition. For elderly care, the project employs a SMART ElderCare approach, utilizing an app to create safer home environments for isolated seniors, thereby reducing risks such as falls. In rehabilitation, Geographic Information System (GIS) software will map community resources to improve discharge planning and ensure patients are cared by the community, not just within the community. Lastly, in community governance, Social Network Analysis (SNA) will be applied to examine social ties and capital in new Macao communities, fostering stronger resident networks. These developments illustrate how AI can move beyond theory into actionable, compassionate support. By embedding technology into social work practice, we can create more responsive, personalized, and sustainable care systems that truly serve vulnerable populations. The presentation would provide details about this ongoing project among three universities in the GRA. A discussion on the approaches to revamping the social work curriculum in relation to AI development will also be covered.

O12.5  Enhancing Social Work Clinical Competence via the BUILD Dialectic Learning Model: A Pilot Study of an Integrated Curriculum

*Renee Chiu¹, Petrus Yat-Nam Ng², Timothy Yuk-Ki Leung¹, Qi-Rong Chen³

¹Department of Social Work, The Chinese University of Hong Kong, China; ²Department of Social Work, Hong Kong Shue Yan University, China; ³Department of Applied Social Sciences, The Hong Kong Polytechnic University, China

Abstract

Background and Purpose: Bridging the gap between theoretical knowledge and practical application remains a persistent challenge in clinical social work education. Traditional lecture-based formats often lack the experiential elements necessary to cultivate the nuanced interpersonal and self-reflective skills required for effective practice. This study evaluates the efficacy of an innovative educational intervention, the BUILD model, which integrates an adaptation of the LEGO Serious Play framework into a clinical social work curriculum. The primary objective is to assess how this tactile, dialectic learning approach enhances undergraduate students' conceptual comprehension, clinical proficiency, and reflective capabilities.

Methods: A mixed-methods, quasi-experimental one-group pre-post-test design was utilized with a convenience sample of undergraduate social work students. The intervention involved an Integrated Curriculum on Clinical Practice (ICCP) comprising nine lectures and six 90-minute bi-weekly integrated tutorials. During these tutorials, students utilized a "BUILD Kit" containing LEGO bricks to visually construct and simulate clinical intervention models based on Person-centered Therapy, Cognitive-behavioral Therapy, and Psychodynamics. Quantitative data were collected at baseline and post-intervention using the Reflective Practice Questionnaire (RPQ) and the Perceived Social Work Competence Scale (PSWCS). Concurrently, qualitative data were gathered through one-on-one, in-depth, semi-structured interviews. Quantitative data were analyzed using paired-sample t-tests, while qualitative transcripts underwent rigorous thematic analysis.

Results: Out of 55 enrolled students, 41 completed the pre- and post-intervention quantitative measures, and five participated in the qualitative interviews. Paired-sample t-tests revealed statistically significant improvements in students' perceived social work competence (t(40) = 2.52, p < .05, Cohen’s d = .39) and reflective practice (t(40) = 2.02, p = .05, Cohen’s d = .32). Notably, significant gains with modest effect sizes were observed in sub-domains including supportive skills (d = .49), therapeutic and insight skills (d = .45), reflective-in-action (d = .45), professional values and ethics (d = .41), and confidence in communication (d = .39). Qualitative findings corroborated these statistical results, highlighting five core themes: perceptible development of clinical skills, robust synthesis of theory and practice, heightened classroom engagement, profound improvement in self-reflective practice, and high appreciation for the innovative teaching modality.

Conclusions and Implications: The integration of the BUILD model represents a promising and viable dialectic innovation in social work education. By utilizing creative, tangible tools to visualize complex theoretical constructs, students can effectively bridge the theory-practice divide, cultivating the empathy, reflexivity, and clinical competence essential for their professional foundations. These findings hold substantial implications for social work pedagogy, suggesting that hands-on, interactive modalities should be increasingly integrated into clinical training curricula to meet the dynamic needs of the field.