Year: 2025
Type: article
Study design: Observational Study
Abstract: The rapid spread of AI tools such as ChatGPT in higher education has elicited interest in their effects on students' learning outcomes. This study investigates students' uses of ChatGPT in relation to university students' academic achievement, and focuses on student engagement, personalized learning experience, and student retention/dropout as potential intervening roles. The study employed a quantitative questionnaire survey of 151 business and management undergraduates in March and April 2025. Respondents responded to a 5-point Likert-scale ordered structure questionnaire tapping 5 constructs: (1) AI use; (2) student engagement; (3) personalized learning experience; (4) student retention/dropout; and (5) academic achievement. Respondents' reported use of ChatGPT clustered in two domains: (a) aiding in STEM-related coursework, research, and analysis, and exam prep; and (b) language learning and translation. In our structural model, ChatGPT use explained 69.3% of the variance in student retention/dropout (R² = .693). The latter, in turn, due to improved focus, persistence, and active engagement, explained 38.5% of the variance in learning motivation (R² = .385). Increased motivation, in turn, related positively to a number of dimensions of academic achievement—creativity and critical thinking, ethical awareness, and knowledge/analysis transfer. Respondents also reported the practical utility of AI tools, time savings, and ease of synthesizing information expeditiously across topics. Exploration analyses reflected low positive correlations between outcome and some national cultural value orientations (e.g., social responsibility awareness, respect for tradition, low uncertainty avoidance, in-group fidelity, long-term persistence, and caring for others). The results, in aggregate, show that students' ChatGPT utilization can serve as a great supplementary resource to raise engagement and retention and have small but positive effects on academic achievement. The findings must be interpreted in terms of the study's cross-sectional research design and limitations in terms of self-reports.
Source: Embedded Selforganising Systems
Author: Khaskhuu, Ariunaa
Institution: Mongolian University of Science and Technology
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: diamond
OpenAlex ID: W7104287482
FWCI:
Cites:
Cited by:
Related to: