AI Use and Thesis Quality among University Students: An Empirical Analysis from Liberia

Abstract

This study examines the relationship between AI usage and thesis quality among university students in Liberia, conducted between May and December 2025. The research utilizes institutional submission report data from 1,059 student documents, which were analyzed using an academic integrity platform to assess three main quality indicators: AI-generated content scores, textual similarity scores (reflecting plagiarism), and grammatical quality scores.
The findings indicate that while AI usage remained relatively low for much of the study period, averaging 20.8%, a statistically significant increase occurred in the final quarter of 2025, resulting in a mean AI score of 52.7% in December. A significant inverse correlation (r = –0.100) between AI scores and similarity scores suggests that students utilizing AI tools tend to produce textually original work. However, the negative correlation between AI usage and grammar scores (r = –0.223) raises important concerns about writing quality and student autonomy.
Undergraduate and graduate students showed a higher dependence on AI compared to the general student population. The article discusses the implications for academic policy, faculty oversight, and the prudent integration of AI methodologies in higher education in Liberia