Demographic factors play a significant role in shaping how students use GenAI. Gender had no statistically significant impact on usage, though males used the tools slightly more across all categories. Grade level, however, was a strong predictor: juniors and seniors were significantly more likely to use GenAI than freshmen and sophomores, especially in course learning, research, and job search. This reflects higher academic demands and possibly greater digital fluency among upper-year students.
A new cross-sectional study reveals how university students in China are leveraging generative artificial intelligence to improve academic performance and support sustainable higher education. Published in Sustainability, the study “University Students’ Usage of Generative Artificial Intelligence for Sustainability: A Cross-Sectional Survey from China,” surveyed 486 undergraduates from Baise University to investigate how GenAI tools are used in real-world educational contexts. The research is grounded in the Technology Acceptance Model (TAM) and Task–Technology Fit (TTF) theory, focusing on the extent, context, and implications of GenAI adoption among students.
The results provide a nuanced picture: while over 79% of students actively use GenAI, familiarity with the tools varies widely, and usage is concentrated mainly in academic tasks such as course learning and research. Meanwhile, the application of GenAI in daily life and job-seeking remains limited, and demographic factors such as major and year of study significantly influence user behavior. The findings underscore the duality of GenAI as both a catalyst for educational innovation and a source of ethical concern, particularly around academic integrity and data privacy.
What is the extent of GenAI tool usage among university students in school?
The study shows that GenAI usage among Chinese university students is widespread but uneven. Nearly four in five students reported using GenAI tools at least occasionally, with 24.5% using them often and 2.3% using them always. Despite this high usage rate, familiarity levels are mixed. Over 43% of students described themselves as neutral or unfamiliar with GenAI, and nearly 10% were completely unfamiliar. These figures suggest a significant gap between access and understanding, indicating that many students are using tools they don’t fully comprehend.
Among the GenAI tools used, Chinese-developed platforms dominate. Doubao and ERNIE Bot are the most frequently used, with usage rates of 78.2% and 66.0% respectively. ChatGPT, while internationally popular, trails behind at 36.2% – a pattern likely influenced by platform restrictions and cultural-linguistic accessibility in mainland China.
In terms of functionality, students gravitate toward tools that simplify common academic tasks. Text generation (91.4%) and information search (81.5%) are the most widely used features, while image generation (42.4%) and translation (30.0%) are moderately popular. Functions like code generation, voice interaction, and grammar checking are used far less frequently. This suggests that students primarily use GenAI for its efficiency in academic writing and information access rather than for more complex or technical functions.
The study also highlights how students learn to use GenAI. More than half engage in active learning, but their primary sources are informal: self-media platforms like social media (83.3%) and knowledge-sharing communities (65.8%). Only a small proportion turn to formal educational settings such as lectures (6.0%) or academic papers (7.4%), revealing a reliance on peer-driven, decentralized learning channels rather than institutional support.
How does GenAI usage vary across different task scenarios?
The study explored four key scenarios: course learning, research activities, daily life, and job search. GenAI usage is most prevalent in course learning, with 87.9% of students using it to complete assignments, 85.4% using it for course-related research, and 76.3% for assignment evaluation. Students also use GenAI to respond to teacher questions and verify academic content, reflecting its deep integration into the learning cycle.
Research activities form the second most common usage scenario. Nearly 79% of students use GenAI for academic writing, and around 67% rely on it to help select research topics. Students also use GenAI to translate academic texts (66.0%) and extract key information from literature (62.1%), suggesting that GenAI is becoming an essential tool for navigating scholarly materials, especially in multilingual contexts.
In daily life, GenAI is used more sparingly but still serves a functional role. Over 71% of students use it to access general knowledge on topics like history and culture. About half use it for personal problem-solving, including diet and financial planning, while smaller groups use it for entertainment or psychological support. This indicates a growing, but still secondary, role of GenAI in non-academic life.
Job-seeking is the least developed scenario. Less than 10% of students use GenAI to generate resumés or simulate interviews, and only 3.5% use it to prepare for job applications. The study attributes this to the limited capabilities of current GenAI tools in tailoring job search content, which often lacks accuracy, personalization, and industry-specific knowledge. Developers are encouraged to improve these functions to enhance relevance and trustworthiness in career support.
What differences exist across gender, grade, and major?
Demographic factors play a significant role in shaping how students use GenAI. Gender had no statistically significant impact on usage, though males used the tools slightly more across all categories. Grade level, however, was a strong predictor: juniors and seniors were significantly more likely to use GenAI than freshmen and sophomores, especially in course learning, research, and job search. This reflects higher academic demands and possibly greater digital fluency among upper-year students.
Disciplinary differences were also pronounced. Students in the arts reported significantly higher usage rates for course learning and research than their counterparts in science, engineering, or agriculture. This is likely due to GenAI’s superior text-processing abilities, which better align with the needs of humanities disciplines. Meanwhile, students in STEM fields may find GenAI’s limitations in logical reasoning, data analysis, and scientific computation to be a barrier, reinforcing the need for further technical refinement.
The study concludes by analyzing open-ended responses from 335 students. The most common recommendation was to introduce formal GenAI courses or lectures to standardize knowledge and raise competency. Students also called for improved accuracy, ethical safeguards to prevent plagiarism, and enhanced personalization in AI outputs. Additional suggestions included expanding functionality, promoting innovation in outputs, and increasing transparency about AI-generated content.
The findings offer a clear call to action for universities and policymakers. Institutions should provide structured education on GenAI, develop tailored training across disciplines, and issue clear ethical guidelines for responsible use. Developers are encouraged to enhance the technical sophistication and cultural adaptability of their tools, particularly for underrepresented tasks like job preparation.
This article is from Devdiscourse