Reuters survey finds all 15 environmental groups use AI despite footprint concerns

All 15 conservation, climate and sustainable-development organizations surveyed by Reuters were using AI, even as they recognized its environmental costs. The finding, described in UkrMedia’s account of the November 7 reporting, captures a practical tension: groups protecting nature are adopting tools whose supporting data centers also require electricity and water.

Their applications include counting animals in aerial photographs, digitizing decades of handwritten research, detecting possible forest fires and monitoring fishing catches. The survey finding concerns those 15 organizations, rather than a measured adoption rate across the environmental sector.

Faster counts—and animals people missed

Kenya’s Save the Elephants offers a concrete example of the reported time savings. A 2025 aerial survey produced about 200,000 photographs. Checking them manually took nine specialists nine months; machine-learning algorithms could perform similar analysis in hours. The group uses models trained with Nvidia graphics technology, according to the Investing.com article published by Yahoo Finance.

The benefit was not just speed. Save the Elephants head Frank Pope said traditional censuses had missed about a third of the elephants, half the zebras and giraffes, and three-quarters of the oryx in the survey area. Population estimates help guide conservation status, funding and habitat protection.

At the Jane Goodall Institute, the challenge is an archive rather than a wildlife count. Handwritten chimpanzee observations from Tanzania include local-language records and scientific abbreviations. Technology developed with Amazon Web Services helps scan, translate and organize that material. The institute estimates the project will take about two years, down from 13 to 15 years using traditional methods.

Other applications put AI closer to an immediate field decision. Conservation International analyzes satellite temperature anomalies in Cambodia to identify possible forest-fire hotspots and alert local conservation patrols.

The resource forecast is bigger than AI

Against those benefits sits a growing infrastructure footprint. The International Energy Agency forecasts global data-center electricity use of about 945 terawatt-hours annually by 2030, more than double the 2024 level. Its water estimate rises from about 560 billion liters in 2023 to 1.2 trillion liters annually by 2030. Electricity powers the servers; water is also needed for cooling.

These forecasts cover all data centers, not AI alone. They describe the scale of the infrastructure issue, rather than the footprint of a wildlife survey or archive project. They therefore cannot, on their own, establish whether any particular conservation application delivers a net environmental benefit.

Keeping the analysis on the boat

The Nature Conservancy’s fishing work shows another approach: changing where the computation happens. AI-enabled cameras on commercial vessels identify fish species and quantities, processing footage onboard instead of sending all the video to remote servers. That reduces the need to transfer large files. It is a specific infrastructure choice within an AI application, rather than a decision to abandon the technology.