Generative AI in Professional Learning at a Chinese Private Vocational College: Student Experiences and Teacher Perspectives

GenAI is now part of ordinary study for many students, but frequent use does not automatically mean deeper learning. This mixed-methods study examined professional-course use of GenAI at a private vocational college in Jiangxi Province, China. The evidence came from 431 student questionnaires and written responses from 18 teachers. After response-quality screening, 184 cases were retained for the primary quantitative analyses; 203 moderately screened cases and the full 431 responses were used for sensitivity checks. Meaningful learning engagement (MLE) was measured with two items and is therefore treated as a brief indicator of cognitive engagement rather than a validated multidimensional scale. Students reported using GenAI most often for knowledge clarification (82.6%) and coursework or exercises (74.5%); practical-training use was lower (20.7%). Perceived professional value (PPV) was associated with MLE (β = .550). When PPV and MLE were entered together, both remained associated with post-adoption learning experience (PAE; β = .339 and β = .364, respectively). The standardized indirect association from PPV to PAE through MLE was .200, with a 95% bootstrap CI of [.109, .307]. These cross-sectional results describe associations rather than causal effects. Teacher accounts added important context. Teachers valued GenAI for quick explanations but also described answer-oriented use, inaccurate professional content, weak links with practical training, and uncertainty about institutional guidance. They wanted more discipline-specific examples, clearer expectations, and assessment that made student reasoning visible. Across both strands, three tensions stood out: efficiency versus engagement, general usefulness versus professional fit, and rapid student adoption versus slower institutional response. Because the evidence came from one college, the findings should be interpreted as context-specific.