The influence of artificial intelligence self-efficacy dimensions on the actual use of DeepSeek among Chinese vocational students

Despite the rapid proliferation of artificial intelligence in vocational education, empirical understanding of the mechanisms driving students’ adoption of generative artificial intelligence remains limited, particularly in the context of DeepSeek among Chinese vocational college students. By extending the Technology Acceptance Model to incorporate artificial intelligence self-efficacy and its sub-dimensions, this study examines key factors influencing students’ actual use of generative artificial intelligence for learning. Data were collected through an online survey administered to 423 students from a Chinese vocational college. Partial least squares structural equation modelling (PLS-SEM) was employed to test the proposed relationships. The results indicate that perceived anthropomorphic interaction, technological skills, and overall artificial intelligence self-efficacy significantly and positively predict students’ actual use of DeepSeek. In contrast, perceived artificial intelligence assistance and comfort with artificial intelligence do not exhibit significant effects. These findings suggest that technical competence and perceived human-like interaction play a more critical role in sustained adoption than general comfort or assistance. This study contributes empirical evidence to artificial intelligence adoption research in vocational education and offers practical implications for enhancing students’ technological competencies and artificial intelligence self-efficacy. Implications and limitations of this study were discussed, and suggestions for future research were also indicated.