As energy and water consumption by artificial intelligence data centres continues to climb, environmental organisations worldwide are confronting a contradiction: they worry about the environmental burden of AI technology, yet increasingly rely on it to protect wildlife, monitor forest fires and conduct environmental research.
A survey of 15 conservation and development organisations found that every respondent has adopted AI to some degree while also expressing concern about its environmental impact. The International Energy Agency (IEA) projects that by 2030, global data centre electricity consumption will reach roughly 945 terawatt-hours, about double current levels, with associated carbon dioxide emissions exceeding 300 million tonnes. Annual water consumption by data centres could also rise from 560 billion litres in 2023 to 1.2 trillion litres.
The United Nations has called for greater transparency regarding the environmental impact of AI systems, and the COP31 climate conference to be held in Turkey this month also plans to push各方 toward commitments on sustainable AI development.
Nevertheless, AI's efficiency advantages in environmental protection are already striking. Kenya-based conservation group Save the Elephants previously needed nine staff members working nine months to analyse roughly 200,000 aerial survey photographs; with AI models, similar work can now be completed in just hours. The Jane Goodall Institute has used Amazon Web Services (AWS) technology to digitise handwritten records from its chimpanzee research project in Tanzania, work originally expected to take 13 to 15 years and now likely to be finished in about two years.
Other conservation bodies are also expanding their use of AI. Conservation International uses satellite imagery and AI to send fire alerts to forest patrol teams in Cambodia; The Nature Conservancy analyses fishing vessel operation footage through shipboard AI cameras, reducing the need to transmit large volumes of data to remote servers. The World Resources Institute controls resource consumption by limiting high-compute image and video generation and reusing existing AI analysis results.
However, attitudes toward AI within the environmental community remain divided. Greenpeace maintains scepticism about continuously expanding data centre infrastructure. How to harness AI to improve environmental protection efficiency while containing its energy, water and carbon emissions is becoming a new challenge for the conservation sector.
Comments