Artificial intelligence (AI), has become instrumental in advancing scientific discovery, industry, and environmental stewardship. They are driving advances in disease diagnosis, renewable energy management, climate prediction, and biodiversity monitoring. Nevertheless, this accelerating rate of AI developments, increased environmental pressures. The rapid creation of larger models and data-centre based infrastructure has placed extreme demands upon critical inputs associated with electricity, fresh water, limited supply of minerals, and depletion of electronic hardware with a relatively short lifetime. AI has contributed to the ecological footprint concerning carbon emissions, water use, supply chain impacts, and electronic waste. Current research on ecological sustainability of AI face challenges due to fragmented data across disciplines. The review aims to gather emerging research emphasizing the sustainability paradox. AI's environmental impact originates from factors like hardware emissions, scaling practices, and rebound effects. Our assessment reveals that its impact will likely exceed available management solutions. Thus, an exclusive focus on reducing efficiency alone will not suffice in the future to minimize environmental impact. In order to create sustainable AI products and systems, changes and transformations must occur within the marketplace, including the development of low-carbon data center infrastructure; the implementation of transparent and accessible reporting; the utilization of environmentally responsible and sensibly sourced computing hardware; and the adoption of circular economy ideals. AI’s ecological future is not predetermined and will ultimately be a product of the cumulative and collective choices regarding technology, policy, and ethics that lead AI development to long-term ecological viability.
扫码关注我们
求助内容:
应助结果提醒方式:
