Generative AI-assisted instruction and college students’ academic self-efficacy, computational thinking, and critical thinking disposition: A six-week pretest-posttest study

Abstract

Purpose – Generative artificial intelligence (AI) is increasingly used in higher education, but evidence about how guided generative AI use influences broad student competencies remains limited. Methods – This study examined a six-week generative AI-assisted instructional program designed around generation, inspection, verification, and independent application, focusing on academic self-efficacy, computational thinking, and critical thinking disposition among 286 college students in Fujian Province, China. A one-group pretest-posttest design was used. Academic self-efficacy was assessed with a 28-item scale, computational thinking with the 29-item Computational Thinking Scales, and critical thinking disposition with the California Critical Thinking Disposition Inventory. Descriptive statistics, reliability analysis, Pearson correlations, and paired-samples t-tests were conducted.Results – All three major outcomes increased significantly. Academic self-efficacy increased from 3.31 to 3.61, computational thinking from 3.28 to 3.57, and critical thinking disposition from 3.36 to 3.63, with moderate paired effect sizes (d = 0.47–0.54).Originality/Value – The study applies a structured four-stage AI-learning process to examine self-efficacy, computational thinking, and critical thinking disposition together rather than treating AI use as a simple exposure condition. Research Implications – The findings support further examination of guided AI use as a form of technology-assisted scaffolding while highlighting the need to distinguish AI-supported learning from non-AI structured scaffolding. Because the study used a one-group design without a control group, the observed changes cannot be attributed to generative AI alone.