Abstract
ABSTRACT
Artificial Intelligence (AI) and Generative AI (GenAI) have become important technologies in the modern world and are widely used in areas such as healthcare, education, banking, transportation, and business. GenAI systems can perform complex tasks such as generating text, images, and other content. However, the growing use of AI and GenAI requires significant computing resources, which can increase electricity consumption and create environmental challenges. Data centres require energy for processing, storage, networking, and cooling, while water may also be used for cooling. These activities can contribute to carbon emissions and other environmental impacts.
This paper explores the relationship between AI energy consumption and environmental sustainability. The study focuses on electricity consumption, carbon emissions, water use, and cooling requirements associated with AI systems. It also examines GenAI models, particularly Transformer-based models and Diffusion Models, to understand their role in energy consumption. The research uses Linear Regression, Decision Tree, KMeans, and Random Forest techniques to analyse energy consumption data, identify patterns, make predictions, and support sustainable decision-making. Based on the findings, the paper proposes sustainable approaches such as renewable energy, energy-efficient AI models, energy-aware scheduling, and smart cooling systems to reduce the environmental impact of AI. The study concludes that AI and GenAI can become more sustainable when energy efficiency and environmental considerations are included in their development and use.