Which educational approaches predict students’ generative AI confidence and responsibility?

Given the risks and ethical concerns of integrating generative artificial intelligence (GenAI) into education, scholars have argued for a critical-reflective education approach that addresses the long-term implications of GenAI. However, empirical research on GenAI education approaches is scarce. This study investigates the prevalence of protective-preventive, critical-reflective, and creative-productive GenAI education approaches in highly digitized Swiss upper secondary schools and how students' experiences of these approaches relate to two aspects of their digital agency: GenAI confidence and GenAI responsibility. Using data from 2357 students, the results showed that the critical-reflective approach was the most commonly experienced and significantly predicted students’ GenAI confidence and responsibility. The creative-productive approach positively and significantly predicted GenAI confidence but not responsibility, while the protective-preventive approach, although the second most common approach, was not significantly related to either outcome. However, these approaches explained little variance in the dependent variables, suggesting that they may not yet be effectively implemented or that digital agency is primarily developed outside schools. Analysis of the control variables showed that identifying as a female had a negative significant effect on GenAI confidence and a positive significant effect on GenAI responsibility. The findings highlight the importance of adopting a critical-reflective GenAI education approach with attention to gender issues in fostering responsible and confident digital citizens.