Reliance on Artificial Intelligence in Chinese Vocational Undergraduate Theses

This study investigates the extent to which vocational undergraduate students in China rely on artificial intelligence (AI) in the process of writing graduation theses, a topic that has gained heightened relevance following the 2025 policy statement by the Chinese Ministry of Education that vocational undergraduates are generally not required to produce such theses. Drawing on comprehensive survey data from MYCOS – a nationally recognized higher education research and consulting organization with a respondent pool of 3,145 teachers and students from diverse institutions, the research analyzes both the prevalence and the patterns of AI-assisted writing, as well as the educational community’s attitudes toward its use. The findings reveal that nearly half of the surveyed teachers (50%) and students (46%) support the cancellation of undergraduate theses, primarily due to their perceived misalignment with vocational training objectives and their often-unsatisfactory quality. Teachers in ordinary undergraduate institutions express stronger support for cancellation than those in elite “Double First-Class” universities, reflecting differing institutional priorities. The study also explores the methods used by educators to detect AI-generated content. The most common strategies include checking for logical and stylistic consistency (64%), oral questioning to verify familiarity with the material (51%), and the use of specialized AI detection tools (41%). However, the accuracy of these tools remains limited, with documented cases of both false positives and undetected AI-polished content. To address the changing educational landscape, the research examines preferred alternatives to traditional theses. Practical projects such as product design, innovation transformation, and entrepreneurial initiatives- emerge as the most favored option (75% of students, 69% of teachers), followed by course-based research reports and vocational skill certifications. In light of these findings, the paper proposes a multi-dimensional framework for responsible AI integration into academic writing. This includes establishing age- and level-specific usage guidelines, embedding critical thinking and verification skills into curricula, and diversifying assessment models to emphasize the reasoning process rather than final outputs alone. Ultimately, the results highlight the need for adaptive, diversified evaluation systems that align with both the evolving capabilities of AI and the labor market’s demand for graduates who combine creative, analytical, and practical competencies.