ChatGPT Use and Critical Thinking Among Undergraduate Students: A Mixed Methods Study

This study examined undergraduate students' academic ChatGPT practices across critical thinking score categories, focusing on verification, use regulation, and perceived dependence. An observational mixed-methods approach combined a critical thinking test comprising sixteen items, a questionnaire comprising fifteen items, and interviews. Pilot testing involved forty students; Cronbach's alpha coefficients were 0.852 and 0.912 for the test and questionnaire, respectively. The presented interview accounts showed variation within and across score categories. High-scoring interviewees described checking information or limiting use, whereas the moderate-scoring interviewee reported accepting answers without verification. The lower-scoring interviewee reported correcting outputs and prompting again. Several interviewees described perceived dependence or reduced independent engagement, but these accounts did not establish cognitive decline. The original Wilcoxon comparisons did not establish an association or causal effect between use and critical thinking. The study's originality lies in distinguishing answer acceptance, iterative prompting, and independent verification rather than assigning uniform practices to score categories. It contributes contextual evidence for investigating responsible AI engagement and evaluating guidance on source checking and independent reasoning, while acknowledging measurement limitations and restricted generalizability beyond participants.