Artificial Intelligence to Consume Water Equivalent to 1.3 Billion People by 2030, UN Warns

) could consume the equivalent amount of water required by 1.3 billion people and an amount of electricity equivalent to over 650 million inhabitants by 2030, according to a report from the UN.The study, conducted by the United Nations University for Water, Environment and Health (UNU-INWEH), warns that the environmental impact of artificial intelligence extends far beyond carbon emissions and that the accelerated growth of this technology could create new pressures on natural resources.According to projections, data centers powering artificial intelligence systems will consume 945 terawatt-hours of electricity by 2030, nearly triple the combined annual consumption of Pakistan, Bangladesh, and Nigeria.Furthermore, their water consumption will be equivalent to the basic needs of the entire population of sub-Saharan Africa, estimated at 1.3 billion people, while the land area occupied by this infrastructure will exceed 14,500 square kilometers. “This report is not a manifesto against artificial intelligence. It is a call to use it responsibly and to address its unintended impacts proactively, to make it sustainable and equitable,” stated Kaveh Madani, Director of UNU-INWEH and lead researcher.The document also warns that, to date, attention has primarily focused on AI’s carbon footprint, neglecting other factors such as water consumption required for cooling data centers and generating energy, as well as the land use needed for technological infrastructure.#### Continue ReadingOne of the most significant findings is that the majority of energy expenditure now comes not from training models, but from their daily use. According to the report, between 80% and 90% of artificial intelligence’s total energy consumption corresponds to the constant processing of user queries.The research indicates that ChatGPT processes approximately 2.5 billion queries per day, representing an estimated annual consumption of 383 gigawatt-hours.Moreover, not all interactions have the same environmental impact. Generating an image using AI can consume up to 1,450 times more energy than a simple text classification, while a video created with this technology can require as much electricity as 200,000 basic spam filtering processes.Finally, the report also alerts about the effects already being observed in various regions. In Ireland, data centers accounted for 21% of electricity consumption measured in 2023, while in Querétaro, Mexico, the expansion of technological infrastructure has put pressure on water resources amidst a prolonged drought. “People think that AI’s environmental footprint decreases as technology improves. But more efficient and affordable AI means more AI consumption, making the total footprint much larger than what we save with efficiency gains,” Madani warned.In light of this scenario, the UN has called upon governments, businesses, investors, and users to consider the environmental impact of artificial intelligence concerning energy, water, and land use, in order to ensure sustainable technological development in the coming years.## You Might Also LikeThe entry