| Title: | A Model for Enhancing Physical Education Teachers’ Self-Efficacy through DeepSeek on Pedagogy Design in Public Universities in Guangxi, China |
| Author: | Lu Haiyong |
| Advisor: | รศ. ดร.จอมพงศ์ มงคลวนิช – Assoc. Prof. Dr. Jomphong Mongkhonvanit |
| Degree: | ปรัชญาดุษฎีบัณฑิต สาขาวิชานวัตกรรมการบริหารการศึกษา – Doctor of Philosophy Program in Educational Administration Innovation |
| Major: | ปร.ด. นวัตกรรมการบริหารการศึกษา – Ph.D. in Educational Administration Innovation |
| Faculty: | บัณฑิตวิทยาลัย (Graduate School) |
| Academic year: | 2569 (2026) |
| Published: | ตีพิมพ์ในวารสาร (Journal Name) Journal of Multidisciplinary in Humanities and Social Sciences Vol. 10 No. 1 pp. 1-19 TCI Tier 1 Publication Year: 2570 (2027) View นำเสนอในที่ประชุมวิชาการ (Conference) The 18th National and International Academic Conference “Sustainable Horizon: Transforming Ideas into Impact” 6-7 August 2025 (pp.253-262) Conference Publication Year: 2568 (2025) View |
Abstract
Artificial Intelligence (AI) is increasingly recognized as a transformative force in higher education; however, its integration into practice-based disciplines such as physical education (PE) remains underexplored. This study developed and validated the DeepSeek-Integrated PE Pedagogy Design Model to explain how DeepSeek adoption—with DeepSeek treated as the focal AI tool in this research—shapes PE teachers’ intention to use DeepSeek in pedagogy design and strengthens their Teaching Self-Efficacy (TSE). Grounded in the Technology Acceptance Model (TAM), the Social Cognitive Theory (SCT), and the Motivation–Opportunity–Ability (MOA) framework, the model specifies that Perceived Usefulness (PU) and Facilitating Conditions (FC) increase teachers’ Intention to Use DeepSeek on Pedagogy Design (ITU), whereas AI Anxiety (AIA) decreases ITU; ITU subsequently enhances TSE, and Perceived Ease of Use (PEOU) moderates these pathways by amplifying positive effects and buffering negative emotional influences.
A sequential mixed-methods approach was employed. Quantitatively, data from 402 PE teachers across 28 public universities in Guangxi were analyzed using Structural Equation Modeling (SmartPLS 4). The constructs demonstrated strong reliability and validity (Cronbach’s α > .85, CR > .90, AVE > .60). The structural results supported the hypothesized relationships: PU and FC significantly increased ITU, AIA exerted a negative influence on ITU, and ITU positively predicted TSE, which served as the model’s central psychological outcome. Four AI-enabled instructional benefits—enhanced task clarity, greater instructional control, stronger classroom confidence, and improved outcome certainty—emerged as key mechanisms through which DeepSeek-supported pedagogy design contributes to strengthened TSE.
Qualitatively, semi-structured interviews with ten experts from Guangxi universities provided contextual validation and interpretation. Thematic analysis (MAXQDA 2020) indicated perceived benefits for lesson planning efficiency, differentiated instruction, data-informed assessment, and reflective practice, while also highlighting concerns about AI-related anxiety and insufficient institutional support.
The study contributes theoretically by extending TAM and SCT into a three-layer framework—Intervention (DeepSeek integration and facilitating conditions), Application (cognitive–affective adoption mechanisms), and Outcomes (AI-enabled instructional benefits leading to TSE). Practically, the findings offer actionable guidance for improving facilitating conditions, enhancing usability, reducing AI anxiety, and embedding DeepSeek-supported pedagogy design into institutional development strategies. The conclusions are tool- and context-specific and should be generalized to other AI tools with caution.
Keywords: DeepSeek; pedagogy design, intention to use, teaching self-efficacy, physical education
