What Drives Vietnamese University Students to Use AI in Learning and Research

A survey of 1,203 university students across Vietnam has found that technology optimism, perceived usefulness, perceived ease of use, technology innovativeness, and perceived enjoyment all exert significant positive effects on students’ satisfaction with artificial intelligence tools and their AI usage behavior in learning and research. The study, published in Discover Education, applied an integrated model drawing on the Technology Readiness Index, the Human-AI Synergy Application Model, and Expectation-Confirmation Theory to explain how students adopt AI for academic tasks such as writing and research support.

The researchers collected data through an online questionnaire distributed from October to November 2025, using a five-point Likert scale. All 1,203 responses collected were deemed valid after screening for incompleteness, duplication, and inconsistent response patterns. The hypotheses were tested using partial least squares structural equation modeling implemented in SmartPLS 4.0, with descriptive analyses and reliability assessments conducted in SPSS 27.0. The study received ethical approval from the Scientific and Training Committee of the National Economics University in Vietnam, and participants were informed that their responses would be kept confidential and used solely for research purposes.

The sample was geographically diverse but showed notable imbalances. Of the respondents, 797 (66.3 percent) were female and 406 (33.7 percent) were male. First-year students made up 43.9 percent of the sample, second-year students 28.7 percent, third-year students 23.0 percent, and fourth-year students 4.4 percent. In terms of residential background, 35.3 percent came from rural areas, 30.6 percent from major cities, 22.9 percent from small cities or towns, 10.6 percent from mountainous regions, and 0.6 percent from coastal or island areas. The authors note that the predominance of female and first-year respondents may introduce bias, and they caution that the findings should be interpreted carefully when generalizing to the broader population of Vietnamese university students.

Most participants reported some familiarity with AI. A majority, 661 students or 54.9 percent, reported a moderate level of experience with AI in learning and research, while 27.8 percent indicated a fairly high level, 5.9 percent a high level, 9.3 percent limited experience, and 2.1 percent very little experience. Regarding frequency of use, 51.9 percent reported moderate use, 29.9 percent frequent use, and 4.2 percent very frequent use, while 11.6 percent reported infrequent use and 2.4 percent very rare use. In terms of daily duration, 34.7 percent used AI for less than 30 minutes per day and 34.4 percent for about one hour, with smaller shares reporting two hours or more.

The structural model tested 14 hypothesized direct relationships, and 12 were supported. Technology optimism showed the strongest associations with perceived usefulness (path coefficient 0.641, p < 0.001) and perceived ease of use (path coefficient 0.378, p < 0.001), indicating that students with positive beliefs about technology tended to perceive AI as more useful and easier to use. Technology optimism also positively predicted technology innovation (path coefficient 0.250, p < 0.001). Perceived usefulness (path coefficient 0.475) and perceived ease of use (path coefficient 0.303) both significantly increased perceived enjoyment, and perceived ease of use positively affected technology innovation (path coefficient 0.309).

Two hypotheses were not supported. Technology insecurity did not significantly affect perceived ease of use (path coefficient 0.019, p = 0.555), the weakest relationship in the model, and technology discomfort did not significantly affect perceived enjoyment (path coefficient 0.028, p = 0.222). The authors suggest that students’ familiarity with technology may mean that feelings of insecurity no longer influence their assessments of how easy tools are to use, and that students today show higher adaptability and are less influenced by technological discomfort when using AI tools.

Notably, the two negative technology readiness characteristics, discomfort and insecurity, showed positive effects on other outcomes. Technology insecurity significantly affected perceived usefulness (path coefficient 0.109, p < 0.001), and technology discomfort significantly affected both perceived usefulness (path coefficient 0.067, p = 0.005) and perceived ease of use (path coefficient 0.278, p < 0.001). The authors interpret these results as suggesting that students who experience discomfort or insecurity may be more aware of the complexity, risks, and limitations of AI tools, and that such awareness may encourage them to seek guidance, improve their AI-related skills, and use AI more carefully and purposefully. They emphasize that these findings should be interpreted with caution and examined further in future research.

Perceived enjoyment emerged as the strongest single predictor in the model, with a path coefficient of 0.573 (p < 0.001) on user satisfaction. The authors state that this finding suggests enjoyable experiences and a sense of pleasure when using AI play a central role in shaping students’ satisfaction, and that emotional factors may complement or even dominate cognitive factors in determining overall satisfaction. Technology innovation also had a positive and significant effect on satisfaction (path coefficient 0.284, p < 0.001), which the authors link to students’ proactive exploration and application of new AI features. The relationship from perceived usefulness to technology innovation was significant in this study (path coefficient 0.085, p = 0.047), although the effect size remained small.

On the measurement side, the authors report that most outer loadings exceeded 0.7 and that the scales showed acceptable internal consistency and convergent validity overall. Some limitations within the measurement model were acknowledged: the Cronbach’s Alpha for the technology innovation construct was 0.541, indicating low reliability, which the authors attribute to the construct being measured with only two indicators and note that Cronbach’s Alpha can be conservative for short scales; the technology discomfort construct yielded 0.690, which the authors describe as acceptable despite falling below 0.7. Several user satisfaction items had lower loadings between 0.581 and 0.666 but were retained because they remained above the minimum threshold of 0.50 and were theoretically relevant. All inner model variance inflation factors were below 5, indicating that multicollinearity was not a concern.

The study is cross-sectional and relies on self-reported data, which the authors note limits the ability to explain changes in students’ AI usage behavior over time. It was conducted mainly among undergraduate students in Vietnam, and the survey used a non-probability convenience sampling approach distributed through academic networks, student groups, and online platforms; respondents were not required to report their specific university, so the exact institutional distribution of the sample could not be identified. The authors recommend longitudinal designs, more balanced sampling, extension to faculty and graduate students, and inclusion of additional variables such as AI literacy, trust, ethical awareness, academic integrity concerns, and institutional support in future work. For institutions, the findings point to practical steps including AI training programs, clear guidelines on responsible and ethical use, and support mechanisms to help students manage discomfort and insecurity when using AI tools in learning and research.

Subject of Research: Factors influencing artificial intelligence use in learning and research among university students in Vietnam

Article Title: Factors influencing artificial intelligence (AI) use in learning and research among university students in Vietnam

Article References: Nguyen, T. B. T., Dao, T. T., Truong, D. A., Hoang, D. T., Bui, C. D., Ngo, M. D., Dinh, D. Q., Tran, Q. Y., Tran, D. T., Duong, D. T., Le, D. G., Vi, T. H., & Truong, D. D. (2026). Factors influencing artificial intelligence (AI) use in learning and research among university students in Vietnam. Discover Education, 5(1), Article 1207. https://doi.org/10.1007/s44217-026-02330-y

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02330-y

Keywords: artificial intelligence, higher education, Vietnam, technology readiness, PLS-SEM, student satisfaction, perceived usefulness, perceived enjoyment, survey research, academic writing, technology acceptance, digital transformation