Green artificial intelligence initiatives: Potentials and challenges

Abstract

Recently, the widespread adoption of artificial intelligence, particularly generative AI technology, has surged across various industries. However, a notable drawback of this technology is its significant energy consumption during model training and operation, which poses challenges to sustainability goals and the environment. Consequently, various initiatives have emerged to promote what is termed "green artificial intelligence," aiming to mitigate these environmental impacts. Nevertheless, research discussing these initiatives remains scarce. Hence, this study aims to identify green artificial intelligence initiatives that contribute to environmental friendliness. This paper has comprehensively reviewed the existing literature, professional websites, and expert blogs to identify and analyze available green AI initiatives. This paper has identified 55 such initiatives, broadly categorized into six themes: cloud optimization, model efficiency, carbon footprinting, sustainability-focused AI development, open-source initiatives, and green AI research and community. This study discusses the strengths and limitations of each initiative to offer a comprehensive overview. The findings provide valuable insights, particularly for industries interested in green artificial intelligence and green technology in general. While some tools have been recognized and studied, comprehensive research and analysis are still required to empirically evaluate the majority of other tools due to their early stages of development in this field.