Environmental groups turn to AI despite concerns over data center emissions

Conservation organizations are embracing artificial intelligence to monitor wildlife, detect wildfires, and analyze environmental research, despite the technology's rising energy and water usage posing challenges to their climate objectives. The conservation sector grapples with a dilemma as data centers supporting AI expand rapidly. According to a recent survey of 15 conservation and development groups, all had incorporated AI to some degree and voiced concerns about its environmental implications.

The International Energy Agency projects data centers to consume approximately 945 terawatt-hours of electricity annually by 2030, a nearly doubling of present-day levels. This surge in electricity consumption is expected to generate carbon dioxide emissions exceeding 300 million metric tons. Additionally, water usage could increase to 1.2 trillion liters per year, up from 560 billion liters in 2023.

These environmental impacts of AI are anticipated to be a key topic at this month's COP31 climate conference in Turkey, where organizers aim to secure commitments for responsible AI development. The United Nations has also called for enhanced transparency regarding the environmental consequences of AI systems.

Efficiency gains in AI applications offer potential solutions to these challenges. In Kenya, Save the Elephants employs machine learning to analyze aerial surveys that previously demanded the effort of nine people over a period of nine months to review around 200,000 photographs. The same process can now be accomplished in a matter of hours using models trained on Nvidia graphics technology.

The Jane Goodall Institute is utilizing Amazon Web Services to digitize handwritten chimpanzee research from Tanzania, a task that would have taken 13 to 15 years under traditional methods and can now be completed in roughly two years. Conservation International leverages satellite imagery and AI to inform forest patrols about potential fires in Cambodia.

The Nature Conservancy processes footage from fishing vessels with onboard AI cameras, diminishing the need to transfer extensive video files to remote servers.

Other groups are striving to minimize unnecessary computing. The World Resources Institute imposes restrictions on intensive image and video generation and reuses previously processed AI results. Greenpeace maintains skepticism towards the expansion of data center infrastructure, reflecting the ongoing debate about whether AI's conservation benefits outweigh its environmental costs.