Environmental Awareness and Perceived Responsibility Regarding Generative AI

Abstract

In recent years, Generative AI (GenAI) has become an integral part of student life. The growing use of GenAI demands a large quantity of resources, e.g., fresh water, electricity, and rare earth metals. This study aims to identify the primary use case for GenAI among undergraduate students at Kyungdong University Global, as well as their awareness of its environmental impact and perceived responsibility. To this end, we examined the interconnections between usage frequency, awareness, responsibility attitude, and readiness to reduce AI use. Students (N=316) were surveyed using a 5-point Likert scale to assess the main use case, awareness, responsibility, and difficulty reducing AI usage. We processed and analyzed the data using the Python Programming Language, with NumPy and Pandas for data manipulation and SciPy for statistical testing. We also utilized scikit-learn for modeling, while Matplotlib for visualization. First, we calculated the mean and standard deviation of the responses to each questionnaire. Then, we classified the responses into seven distinct categories: Total AI Usage Score (TUS), Average AI Usage Score (AUS), Number of GenAI Use Categories (AUC), General Environmental Concern Score (GCS), GenAI Environmental Awareness Score (AAS), Responsible AI Attitude Score (RAS), and AI Reduction Feasibility Score (RFS). For AAS and RAS Cronbach’s α= 0.761 and 0.800 respectively. The AUC values showed limited variation since students selected only one primary use category. Therefore, it was not useful for explaining differences in AAS and RAS. The majority of students had a moderate to high level of awareness regarding the environmental impact of GenAI (57.3% highly aware). Two regression models were examined for AAS and RAS. The first regression suggested that AAS is associated with both TUS(β=0.139,p=0.004) and GCS(β=0.138,p=0.001). The second regression suggested that RAS had a statistically significant association with AAS(β=0.211,p<0.001) and GCS(β=0.207,p<0.001). Spearman correlation further revealed that TUS was positively associated with AAS (ρ=0.202,p<0.001) with RAS (ρ=0.126,p<0.05), but negatively associated with RFS (ρ=-0.148,p<0.01). The analysis showed that the students who use GenAI frequently are more aware of its environmental impact but find it hard to stop using it. Additionally, the results show that simply having environmental awareness is insufficient; students require alternative solutions to address this issue.