Using AI could have a hidden cost: Water needed by 1.3 billion people

As artificial intelligence continues to integrate into daily life, facilitating faster work, learning, and content creation, a new report from the United Nations University (UNU) raises alarming concerns regarding its environmental footprint. The study highlights the significant demand for water and energy associated with the rapid growth of AI technologies, warning that this could lead to severe ecological challenges in the near future. By the year 2030, data centers essential for powering AI systems may consume enough water to satisfy the basic domestic needs of approximately 1.3 billion individuals, a number comparable to the entire population of Africa. This staggering statistic underscores the often-overlooked water consumption linked to AI, overshadowed by the more frequently discussed carbon emissions. The UNU report indicates that substantial water is required for cooling servers and generating the electricity necessary for operating advanced AI models. Furthermore, it forecasts that AI-driven data centers could use a staggering 945 terawatt-hours of electricity annually by 2030, nearly three times the total yearly electricity consumption of Pakistan, Bangladesh, and Nigeria combined. This increasing demand for computing power is placing immense strain on global energy systems. The study emphasizes that AI is not merely a digital phenomenon; it is supported by a complex physical infrastructure comprising data centers, sophisticated chips, cooling systems, and energy networks. Constructing and maintaining this infrastructure demands vast resources, including land, water, and electricity. One particularly striking revelation from the report is that everyday interactions with AI may consume more energy than the initial training of AI models. Researchers estimate that routine usage accounts for about 80 to 90 percent of the total energy demand, with tasks like image and video generation requiring significantly more energy than simple text-based requests. In light of these findings, the United Nations University and the UN advocate for a balanced approach to AI development. They emphasize the importance of considering the environmental impact of water, land, and energy in conjunction with technological advancement, urging governments, tech companies, and users to embrace responsible practices as AI continues to evolve.