Generative AI, Critical Thinking, and Human Agency A Computational Study of Decision-Making in Digital Environments

This study explored the connections between use of generative artificial intelligence (GenAI), critical thinking, human agency, cognitive dependence, trust in AI, and academic decision-making in a digital learning environment. The method used was a quantitative cross-sectional design, based on a survey of 2,613 university students and 30 variables related to demographic characteristics and five behavioral constructs. The study used descriptive statistics, Cronbach's alpha, Pearson's correlation, multiple linear regression, and predictive machine-learning models to analyze composite scores. The results revealed a generally high level of GenAI use in academic activities, with a mean score of 3.80, and critical thinking responses and AI-assisted academic decision-making both scoring 3.75. Academic decision-making was strongly associated with cognitive dependence and trust in AI, while there was a lesser bivariate association with reduced critical thinking. The overall variance in academic decision-making was accounted for by 40.8% of the variance in the multiple regression, with standardized cognitive dependence being the strongest predictor followed by AI usage and trust. Moderate discrimination was also observed for predictive classification, with the best ROC-AUC score of 0.748 for XGBoost. The overall findings suggest a growing integration of GenAI into students' academic reasoning and decision-making, with increased reliance potentially compromising autonomous cognitive engagement and agency-related behaviors. The study points to the importance of responsible AI literacy, verification practices and learning designs, which guarantee critical thinking and autonomous decision-making.

Hali tarjima qilinmagan