Generative AI Use and Burnout Risk in Higher Education: A Cross-Sectional Analysis of 50,000 Students

The rapid adoption of generative artificial intelligence (GenAI) in higher education has raised concerns about student well-being, dependency, and skill retention. This study examined the associations between GenAI use, perceived dependency, academic context, burnout risk, and skill retention. A secondary cross-sectional analysis of 50,000 students was conducted using descriptive analysis, machine-learning models, and behavioural clustering. High burnout was reported by 25.0% of students and was most strongly associated with weekly GenAI use, perceived dependency, and exam-related anxiety. A distinct cluster of intensive GenAI users showed a High-burnout rate of 62.3% and lower mean skill retention than the remaining clusters. The binary XGBoost model achieved a ROC-AUC of 0.802, the three-class model achieved a macro F1 score of 0.538, and the skill-retention model explained 13.6% of the variance. The findings suggest that intensive GenAI use and perceived dependency are associated with increased burnout risk, while skill retention reflects a broader combination of behavioural and academic factors, although no causal conclusions can be drawn.