Artificial intelligence is transforming the way we study wildlife. In Kenya, algorithms have learnt to analyse hundreds of thousands of photographs of animals in a matter of hours \u2014 a task that used to take months. At the same time, scientists are using AI to study chimpanzee behaviour, detect forest fires and monitor industrial fishing. However, the very technology that helps protect the environment is itself consuming ever-increasing amounts of electricity and water. This is according to a Reuters study published on 7 November, on the eve of the COP31 climate summit.<\/p>\n
The Kenyan conservation organisation Save the Elephants uses machine learning to count wild animals from aerial photographs. During a large-scale survey of a conservation area in 2025, researchers obtained approximately 200,000 photographs.<\/p>\n
It took nine specialists nine months to check these images manually. Now, algorithms trained using graphics processing units carry out the same work Nvidia<\/a>, can be completed in a matter of hours.<\/p>\n The results were unexpected. According to Frank Pope, head of Save the Elephants, traditional censuses had failed to account for around a third of the elephants, half of the zebras and giraffes, and three-quarters of the oryx \u2014 antelopes that inhabit the arid regions of Africa.<\/p>\n Such a difference has practical significance for nature conservation. Based on estimates of animal numbers, the status of populations is determined, funds are allocated to nature conservation programmes, and measures to preserve their habitats are planned. Inaccurate counts can lead to an underestimation of the scale of the problem or the misallocation of resources.<\/p>\n Reuters cites another example from Tanzania, where the Institute Jane Goodall<\/strong> is working on digitising many years\u2019 worth of field observations of chimpanzees in Gombe National Park.<\/p>\n For decades, researchers recorded the behaviour of primates by hand, using the local Kicha language and special scientific abbreviations. Due to the complexity of the material, it could take between 13 and 15 years to process the archive using traditional methods.<\/p>\n Using technology developed in collaboration with Amazon Web Services, records can be scanned, translated and organised much more quickly. According to Liliana Pintea, Vice-President of the Institute for Environmental Research, the overall duration of the project has been reduced to approximately two years.<\/p>\n Digitised archives will help researchers track changes in chimpanzee behaviour over decades, compare historical observations with current data, and better assess the impact of environmental changes on populations.<\/p>\n Artificial intelligence is not only used to study animals. Conservation International has begun analysing satellite data on temperature anomalies in Cambodia. Algorithms help to identify potential forest fire hotspots and send alerts to local nature conservation patrols.<\/p>\n The American organisation The Nature Conservancy takes a different approach. It installs AI-enabled cameras on commercial fishing vessels to identify the species and quantities of fish caught.<\/p>\n Processing video directly on-board reduces the need to transfer large volumes of footage to remote servers. This not only speeds up analysis, but also helps to reduce energy consumption and the load on the digital infrastructure.<\/p>\n Despite the benefits for environmental organisations, the widespread adoption of artificial intelligence is creating new environmental problems. Data centres require significant amounts of electricity to power their servers and water to cool them.<\/p>\n According to a forecast by the International Energy Agency (IEA), global electricity consumption by data centres could exceed 945 terawatt-hours per year by 2030. This is more than double the 2024 level and roughly equivalent to Japan\u2019s annual electricity consumption.<\/p>\nJane Goodall\u2019s research on chimpanzees is being digitised using AI<\/h2>\n
Algorithms are already detecting fires and monitoring industrial fishing<\/h2>\n
By 2030, data centres could consume more electricity than the whole of Japan<\/h2>\n