Why University Students Keep Using Generative AI Starts With Confirming Their Expectations Were Met
An analysis of 282 university students finds that expectation confirmation sharply raises satisfaction, and satisfaction, usefulness, and social influence drive continued use, while anxiety about AI has no effect — published in the international journal Sustainability
A study analyzing why university students keep using generative AI has been published in the international journal Sustainability (SSCI·SCIE). The collaborative study was led by first author Young Mee Jung of the Department of Aviation Tourism at Hanseo University, with researchers at the HJ Institute of Technology and Management (HJITM) serving as corresponding author.
The researchers analyzed responses from 282 university students who use generative AI with structural equation modeling. Building on the expectation-confirmation model (ECM) and the technology acceptance model (TAM), they examined whether knowledge application (the degree to which students actually apply knowledge gained from AI) and perceived intelligence of AI lead through usefulness and 'expectation confirmation' to satisfaction and continuance intention, and what roles are played by social influence, such as encouragement from those nearby, and anxiety arising from AI's appearance and configuration.
The analysis showed that perceived intelligence raised both expectation confirmation (β=0.547) and usefulness (β=0.271), and knowledge application raised both usefulness (β=0.275) and expectation confirmation (β=0.274). Expectation confirmation in particular raised satisfaction very strongly (β=0.752) and also raised usefulness (β=0.394). In other words, the experience of finding that the tool delivered as expected is the key determinant of satisfaction.
Satisfaction (β=0.347), usefulness (β=0.303), and social influence (β=0.264) all significantly affected continuance intention. By contrast, fear and discomfort stemming from AI's appearance or configuration had no significant effect on continuance intention (β=−0.050). The model explained 64.1% of university students' intention to continue using generative AI.
The study is notable for empirically identifying the conditions under which generative AI is used consistently in education rather than as a one-off experiment. It suggests that universities and service developers can be effective by designing usage experiences in which students can directly confirm the results they expected in assignments and learning, and by creating environments where peers and instructors share how they use the tools.
HJITM continues to conduct joint research with scholars in Korea and abroad on the use of generative AI in education.
Key findings
- Perceived intelligence → expectation confirmation (β=0.547), usefulness (β=0.271); knowledge application → usefulness (β=0.275), expectation confirmation (β=0.274)
- Expectation confirmation → satisfaction (β=0.752), the strongest effect
- Satisfaction (β=0.347), usefulness (β=0.303), social influence (β=0.264) → continuance intention
- Anxiety about AI appearance and configuration has no effect · explained variance 64.1% · N=282
- Paper
- Understanding Continuance Intention of Generative AI in Education: An ECM-Based Study for Sustainable Learning Engagement
- Authors
- Young Mee Jung, Hyeon Jo
- Journal
- Sustainability (MDPI, SSCI·SCIE), Vol. 17, 6082, 2025
- DOI
- 10.3390/su17136082
- Cite
- Young Mee Jung, Hyeon Jo (2025). Understanding Continuance Intention of Generative AI in Education: An ECM-Based Study for Sustainable Learning Engagement. Sustainability, 17, 6082.
This story could become your next paper
HJITM works with researchers from study design through surveys, analysis and submission to international journals in generative AI and technology management. Bring us your research question.