Study shows students had low and often ineffective reliance on ChatGPT-4 in quizzes, with persistent misuse patterns and clear behavioral predictors, highlighting the need for better training and smarter AI design to support effective and ethical use.
