Clustering generative AI usage among university students based on productivity, verification, and integrity

Abstract

Purpose: to identify and classify university students’ generative AI usage patterns based on three dimensions: academic productivity, information verification, and academic integrity.

Method: this study employed a quantitative cross-sectional survey involving 200 university students. Data were collected using a Likert-scale questionnaire measuring academic productivity, information verification, and academic integrity, and analyzed using K-Means clustering with the Elbow Method, Silhouette Score, and Davies–Bouldin Index.

Findings: identified three distinct generative AI usage profiles among university students: limited-ethical users (15.0%), high-risk pragmatic users (40.0%), and productive-critical-ethical users (45.0%). The findings indicate that the largest group successfully combined high academic productivity, strong information verification, and high academic integrity, while a substantial proportion of students demonstrated high AI utilization with insufficient verification and ethical awareness.

Implications: for higher education institutions to develop targeted AI literacy programs, verification skills training, and responsible AI use policies based on student behavioral profiles.

Originality: lies in the development of a behavioral classification model in the use of generative AI that integrates the dimensions of academic productivity, information verification, and academic integrity through a clustering approach.