Sarcouncil Journal of Engineering and Computer Sciences

Abstract: The rapid advancement of artificial intelligence has created an unprecedented environmental challenge that threatens to undermine the sustainability of technological progress. This article examines the complex relationship between AI development and environmental impact, revealing how the computational demands of modern AI systems, particularly large language models and vision transformers, generate carbon emissions that rival traditional industrial sources. The article explores the paradoxical nature of AI's environmental footprint, where systems designed to address global challenges contribute significantly to climate change through their enormous energy consumption during training, deployment, and inference phases. The article investigates emerging Green AI solutions, including energy-efficient architectures, knowledge distillation, carbon-aware scheduling, and renewable-powered infrastructure, which offer pathways toward more sustainable AI development. Furthermore, it analyzes the intricate balance between model performance and environmental impact, demonstrating how organizations can implement carbon budgeting and efficiency-aware strategies to reduce emissions while maintaining competitive performance. The ethical dimensions of AI sustainability are critically examined, highlighting the disproportionate burden placed on climate-vulnerable communities in the Global South, who contribute least to AI's carbon footprint yet suffer most from its environmental consequences. This comprehensive article underscores the urgent need for a fundamental shift in AI development priorities, advocating for inclusive governance frameworks and collaborative international approaches that ensure technological advancement does not exacerbate existing environmental inequalities or create new forms of climate injustice.