WASHINGTON, UNITED STATES — Artificial intelligence is helping environmental organizations detect wildfires, monitor endangered animals and analyze decades of scientific research at unprecedented speeds.
But the technology is creating an uncomfortable dilemma for the very organizations fighting to protect the planet.
The AI systems helping conservationists save forests and wildlife also depend on data centres that consume electricity, require cooling and contribute to greenhouse-gas emissions.
A Reuters investigation published on November 7 and republished by Channel News Asia found that all 15 environmental, conservation and development organizations interviewed were using artificial intelligence in some form.
Every organization also expressed concern about the environmental consequences of the infrastructure supporting it.
The findings expose a growing contradiction in the global AI revolution.
Conservation groups are discovering that machine learning can accomplish scientific tasks that previously required months or even years.
Yet the computing infrastructure behind these breakthroughs is expanding so rapidly that researchers expect global data-centre electricity consumption to roughly double by 2030.
Some environmental organizations are now embracing AI while demanding stricter environmental safeguards.
Others, including Greenpeace, are questioning whether the enormous construction boom is necessary in the first place.
As governments prepare to discuss sustainable artificial intelligence at the COP31 climate conference in Turkey, a fundamental question is becoming increasingly urgent.
Can AI become one of the world’s most valuable environmental tools without creating a new environmental burden?
AI Helps Researchers Discover Wildlife That Human Surveys Missed
One of the most remarkable examples comes from Kenya.
Save the Elephants, a conservation organization working with the Wildlife Conservation Network, is using machine learning to improve wildlife surveys and analyze elephant behaviour.
Traditionally, researchers relied on people examining aerial photographs to identify and count animals.
The process was extraordinarily time-consuming.
According to Save the Elephants chief executive Frank Pope, an aerial survey of Kenya’s largest protected area in 2025 produced approximately 200,000 photographs.
Nine people spent nine months manually reviewing the images.
Now, machine-learning models trained using Nvidia GeForce graphics platforms can complete comparable image analyses within hours.
But the improvement is not simply about speed.
The organization’s AI-assisted assessments indicated that earlier human surveys had overlooked substantial numbers of animals.
Researchers reported that the previous methods missed approximately one-third of elephants, half of zebras and giraffes, and three-quarters of oryx.
These are findings reported by Save the Elephants from its survey comparisons, not universal estimates of wildlife-counting errors.
Nevertheless, the results highlight how machine learning may improve conservation planning.
Accurate population counts can help researchers identify threats, assess habitat conditions and make stronger cases for conservation funding.
For endangered species, better information could influence how limited protection resources are allocated.
Why Artificial Intelligence Is Changing Conservation Science
Wildlife conservation generates enormous quantities of information.
Camera traps capture photographs and videos.
Drones survey remote habitats.
Satellites observe changes in forests and vegetation.
Researchers record animal movements, migration patterns and behaviour.
But analyzing this material manually can require extensive time and personnel.
Artificial intelligence can identify patterns across large datasets and help researchers focus on observations requiring closer examination.
Image-recognition systems can distinguish animals from their surroundings.
Machine-learning models can help classify species or flag unusual changes.
Other AI applications can process satellite imagery to identify potential environmental threats.
However, these systems are not automatically accurate.
Their performance depends on training data, environmental conditions and independent validation.
Human researchers remain essential for checking results and interpreting their significance.
For conservation organizations, the technology’s value lies in helping specialists make better use of information they already collect.
Conservation International Uses AI to Detect Wildfire Threats in Cambodia
Artificial intelligence is also being deployed to identify potential forest fires.
Conservation International has introduced AI-assisted monitoring of satellite-detected hotspots in environmentally sensitive areas of Cambodia.
The system evaluates satellite information and sends alerts to community forest patrols when conditions suggest a possible wildfire threat.
The objective is to reduce the delay between detecting a problem and responding to it.
In remote forests, early warning can be especially valuable.
Traditional patrols may struggle to monitor large areas continuously.
Satellite imagery can provide broader coverage, while AI can help prioritize alerts.
But identifying a hotspot is not the same as confirming an active wildfire.
Alerts still require appropriate interpretation and, when necessary, field verification.
Ali Swanson, Conservation International’s AI lead, told Reuters that the organization recognizes the environmental costs associated with artificial intelligence.
She nevertheless argued that the urgency of climate change and biodiversity loss creates a strong case for using tools that can accelerate conservation work.
The challenge is determining when the environmental benefits justify the resources consumed.
Jane Goodall’s Decades of Chimpanzee Research Get an AI Breakthrough
Another extraordinary application involves the scientific legacy of Jane Goodall.
The Jane Goodall Institute is using artificial intelligence to digitize decades of handwritten research notes documenting chimpanzee behaviour in Tanzania’s Gombe National Park.
The records contain detailed observations accumulated over many years.
Processing them manually was expected to require approximately 13 to 15 years.
The challenge was particularly demanding because the notes included handwriting, the local Kiha language and scientific shorthand.
With a tool developed in collaboration with Amazon Web Services, researchers can scan, translate and organize the material much more quickly.
The institute now expects the overall project to take approximately two years.
That is a significant reduction in the anticipated timeline.
Lilian Pintea, the institute’s vice president of conservation science, emphasized the importance of keeping people in control of how these technologies are applied.
The project illustrates the potential for AI to make historical scientific observations more accessible.
Digitizing the records could enable researchers to examine long-term changes in chimpanzee behaviour and environmental conditions.
But AI-generated transcriptions and translations still require quality checks.
The goal is to preserve and use the original research more effectively—not replace scientific judgment.
The Environmental Contradiction: AI Needs Enormous Computing Power
These conservation achievements come with an important complication.
Artificial intelligence generally requires computing infrastructure.
Developing, training and operating AI models can involve processors housed in data centres.
Those facilities need electricity, cooling equipment and supporting infrastructure.
Their environmental impacts depend on several factors, including the electricity-generation mix, equipment efficiency, cooling technology and location.
The International Energy Agency estimated that data centres consumed approximately 415 terawatt-hours of electricity globally in 2024.
That represented around 1.5% of worldwide electricity demand.
Its 2025 Energy and AI report projected consumption reaching approximately 945 terawatt-hours by 2030.
A subsequent IEA update placed the central projection closer to 950 terawatt-hours.
Both estimates indicate that demand could roughly double over the decade.
However, these figures cover data centres generally.
They include conventional computing, cloud services and other digital workloads—not artificial intelligence alone.
AI is an important driver of increasing demand, but it is not responsible for every unit of electricity consumed by data centres.
The challenge is that rapid expansion may strain local electricity grids even when the industry’s global share remains comparatively modest.
Data-Centre Electricity Demand Could Rival Japan’s Annual Consumption
The IEA’s projections place the scale of the infrastructure expansion into perspective.
Approximately 950 terawatt-hours would be comparable to Japan’s annual electricity consumption.
Data centres could account for around 3% of global electricity demand by 2030 under the updated central projection.
But the impacts will not be distributed evenly.
The United States and China are expected to account for much of the growth.
Some regions may experience particularly concentrated demand because large computing facilities are clustered together.
That can create challenges for grid operators.
Electricity systems must provide reliable power while also supporting homes, factories and transport infrastructure.
In some locations, building new data centres may require additional substations, transmission lines and generating capacity.
The environmental implications depend on how that electricity is produced.
A facility relying largely on fossil-fuel generation has a different emissions profile from one supplied by low-carbon energy.
But renewable-energy purchases do not automatically eliminate every environmental impact associated with electricity use.
Grid constraints, construction materials and cooling requirements must also be considered.
AI’s Water Consumption Is Becoming Another Major Concern
Electricity is only part of the environmental equation.
Some data centres use water to remove heat from servers and cooling equipment.
Reuters cited International Energy Agency projections suggesting that global data-centre water consumption could rise from approximately 560 billion litres in 2023 to 1.2 trillion litres annually by 2030.
The figures represent estimates and projections.
They should not be interpreted as direct measurements of every facility.
Water requirements can differ dramatically depending on cooling technology, climate and energy supply.
Some facilities use evaporative cooling.
Others employ air cooling or closed-loop liquid systems that reduce direct water consumption.
The water footprint of electricity generation can also be relevant when assessing broader impacts.
For communities already experiencing drought, even a relatively small additional demand may become controversial.
That has led environmental groups to demand greater transparency from technology companies.
The crucial question is not simply how much water the global industry uses.
It is whether individual projects can operate without creating unacceptable pressure on local resources.
Communities Are Already Fighting Back Against Massive Data Centres
The conflict between technological investment and environmental protection is becoming increasingly visible.
Across Europe, Asia and Africa, residents and campaigners are questioning proposed data-centre developments.
Their concerns include electricity demand, water use, construction disruption and the effects on nearby communities.
As previously reported by WWC ONE MEDIA in its investigation into the global AI data-centre backlash, environmental and permitting disputes have affected major proposed investments.
Some projects face delays, restrictions or opposition.
But investments affected by disputes should not automatically be described as money permanently lost.
Projects may eventually proceed after changes to their design, location or approval conditions.
The broader issue is that communities increasingly want to understand who benefits from large AI developments and who bears their environmental costs.
A facility may bring construction jobs, business investment and tax revenue.
Yet its electricity and water requirements can create additional pressure on local infrastructure.
Governments must assess those trade-offs before approving major developments.
Greenpeace Questions Whether the AI Building Boom Is Necessary
Greenpeace International represents one of the more skeptical perspectives.
The organization told Reuters that it uses some AI tools, partly because AI features are already incorporated into software it relies on.
However, Greenpeace is developing internal guidelines emphasizing transparency, human oversight and deliberate use.
It also continues challenging the large-scale construction of data centres.
Katrin Xueying Wu, a climate and energy campaigner with Greenpeace, questioned whether every proposed expansion of computing infrastructure is genuinely necessary.
Her concern highlights a distinction between useful applications and demand generated simply because AI technology is available.
For example, using machine learning to identify an endangered animal may have a clear conservation purpose.
Generating large quantities of unnecessary content could create computing demand without comparable public benefit.
That does not mean all generative AI applications are inherently wasteful.
Their value depends on what they accomplish and the resources they require.
The broader Greenpeace argument is that responsible AI policy should examine whether demand can be reduced, not only whether additional electricity can be supplied.
Some Environmental Groups Are Finding Ways to Reduce AI’s Footprint
Several organizations are already taking practical steps to limit resource-intensive computing.
The World Resources Institute, which uses AI for environmental and climate analysis, told Reuters that it restricts certain demanding applications, including unnecessary image and video generation.
It also caches AI query results.
That means previously completed requests can sometimes be reused rather than processed repeatedly.
The organization hopes such practices will reduce computing requirements.
The Nature Conservancy is exploring another approach.
It uses AI-assisted video analysis aboard commercial fishing vessels to assess catches and identify different species.
Instead of transferring enormous quantities of footage to remote data centres, it deploys AI-enabled cameras capable of processing information on the vessels themselves.
This approach is known as edge computing.
It can reduce data transfers and may lower the computing and communications resources required for particular tasks.
Vienna Saccomanno, a senior scientist at The Nature Conservancy, told Reuters that the engineering effort was worthwhile because of the potential environmental benefits and improved information.
However, edge computing is not automatically more energy-efficient in every situation.
Its advantages depend on equipment design, processing requirements and the amount of data being transferred.
For conservation organizations, selecting the right technical approach can be an important part of responsible AI use.
BirdLife International Says Refusing AI Could Carry Its Own Risks
Other conservation organizations take a more enthusiastic view.
BirdLife International sees artificial intelligence as a potentially important tool for protecting biodiversity.
Chief scientist Stuart Butchart told Reuters that conservation groups have an obligation to consider technologies capable of improving their work.
His reasoning is practical.
Environmental organizations often operate with limited funding and personnel.
They must monitor large geographic areas and respond to threats involving habitat loss, climate change and wildlife decline.
If AI allows researchers to process information faster and direct resources more effectively, refusing the technology could carry an opportunity cost.
Butchart’s argument does not dismiss AI’s environmental footprint.
Instead, it suggests that the benefits of some applications may outweigh their costs.
The difficulty is measuring that trade-off accurately.
A model that improves wildlife surveys could deliver meaningful conservation benefits.
But the computing resources needed to operate it still have environmental consequences.
Determining the net impact requires evidence about both sides of the equation.
COP31 Could Put Sustainable AI Under Global Scrutiny
The controversy is expected to become part of discussions surrounding COP31, the United Nations climate conference being hosted by Turkey in November 2026.
Reuters reported that Turkey intends to encourage governments, businesses and institutions to support principles for sustainable AI development.
UN Secretary-General António Guterres has also called for major AI companies to disclose the environmental impacts of their systems.
Such disclosure could help researchers, regulators and consumers evaluate competing sustainability claims.
Important measures include electricity consumption, greenhouse-gas emissions, water use and the environmental effects of computing infrastructure.
Transparency matters because different AI systems can require very different amounts of computing.
A small image-classification model may operate with relatively limited resources.
A large general-purpose model can require significantly more infrastructure.
But comparing systems is difficult without consistent measurement and reporting.
There is also a distinction between proposed voluntary commitments and legally enforceable regulations.
The COP31 discussions may encourage stronger standards, but they do not by themselves establish a binding worldwide AI environmental law.
Big Technology Companies Face Their Own Climate Dilemma
The environmental debate extends to some of the world’s largest corporations.
Microsoft, Amazon and Google’s parent Alphabet have reported emissions pressures linked partly to expanding data-centre infrastructure.
As AI demand increases, technology companies must reconcile their growth plans with long-term climate commitments.
Amazon told Reuters that its AI products were among the more sustainable offerings because of low-carbon energy purchases and efforts to improve water and electricity efficiency.
That is the company’s assessment rather than an independently established ranking of every AI provider.
Reuters said Microsoft and Google did not respond to its requests for comment for this investigation.
The broader issue is that improvements in efficiency may not be enough if total demand grows faster.
A data centre can become more efficient per unit of computing while consuming more electricity overall as workloads expand.
This phenomenon is sometimes discussed in relation to rebound effects, where efficiency gains can be partly offset by greater usage.
For technology companies, credible sustainability strategies must therefore consider total resource consumption alongside efficiency improvements.
China’s AI Infrastructure Boom Shows How Difficult the Balance Can Become
China is also confronting the relationship between computing growth and environmental resources.
In Inner Mongolia, Ulanqab has emerged as an important computing hub because of relatively inexpensive electricity, available land and its cooler climate.
Yet the region has a history of drought.
As WWC ONE MEDIA previously reported on China’s expanding AI data-centre infrastructure in Inner Mongolia, future water availability could become an important concern.
That does not mean existing data centres have already caused a regional water crisis.
It means rapid development requires careful planning around long-term resource availability.
The situation illustrates why environmental impact cannot be evaluated using electricity prices alone.
Land, grid infrastructure, water supply and local economic conditions all matter.
Southeast Asia Is Becoming a Major Test for Sustainable AI
The debate is especially relevant to Singapore and Malaysia.
The International Energy Agency expects data-centre electricity demand in Southeast Asia to more than double by 2030, partly because of expansion around Singapore and southern Malaysia.
Johor has emerged as one of the region’s fastest-growing digital infrastructure markets.
But the construction boom is already generating community concerns.
As WWC ONE MEDIA reported on the impact of Johor’s data-centre expansion on Iskandar Puteri residents, some homeowners have raised questions about nearby infrastructure, construction disruption and their properties.
Those concerns do not establish that every data centre causes environmental harm or reduces property values.
But they demonstrate why local communities want greater involvement in development decisions.
For Southeast Asian governments, the challenge is attracting investment without placing excessive pressure on electricity grids, water systems and surrounding neighbourhoods.
US Data-Centre Expansion Faces New Questions Over Public Land
The same tensions are emerging in the United States.
Proposed AI developments on federally managed land are attracting scrutiny from lawmakers.
As WWC ONE MEDIA reported in its coverage of Senator Martin Heinrich’s questions about proposed projects covering more than 17,000 acres, the debate includes concerns about land use, water demand, electricity infrastructure and public accountability.
The proposed projects are at different stages and should not be treated as completed facilities.
Their environmental effects will depend on the developments ultimately approved and built.
Nevertheless, the controversy illustrates why AI infrastructure is becoming a political issue.
Governments want technological competitiveness and private investment.
Communities want environmental protection and affordable essential services.
Balancing those interests will remain difficult as demand for computing power increases.
What Would Truly Sustainable AI Look Like?
The environmental organizations interviewed by Reuters are beginning to identify possible approaches.
One involves using smaller or more efficient AI systems when they can accomplish the required task.
Another is limiting unnecessary high-compute applications.
A third is processing data closer to where it is collected when that reduces overall resource use.
Organizations can also reuse previous computational results instead of running the same analysis repeatedly.
For infrastructure operators, important measures may include improved cooling, low-carbon electricity, transparent resource reporting and careful site selection.
But no single technique guarantees sustainability.
A cooling system that reduces water consumption may increase electricity requirements.
A facility using renewable electricity may still depend on backup generators or resource-intensive construction materials.
And a highly efficient model may still have a large total footprint if it is used millions of times.
Sustainability therefore requires examining the full lifecycle of computing systems.
That includes hardware manufacturing, operation, electricity generation, cooling and eventual equipment disposal.
The Biggest Question: Can AI’s Environmental Benefits Outweigh Its Costs?
The issue ultimately comes down to whether artificial intelligence delivers enough measurable value to justify the resources it consumes.
In some conservation applications, the benefits appear compelling.
Save the Elephants has reported major improvements in wildlife counting.
The Jane Goodall Institute expects AI to help reduce a lengthy digitization project from as much as 15 years to approximately two.
Conservation International is using AI to support early wildfire alerts.
The Nature Conservancy is applying machine learning to fisheries monitoring.
These are specific applications with identifiable purposes.
But their success does not automatically justify every large AI investment or data-centre development.
Environmental groups still need to assess how much computing their applications require.
Technology companies need to provide credible information about energy use, water consumption and emissions.
Governments must evaluate whether new infrastructure can operate sustainably within local constraints.
And researchers must continue testing whether AI actually improves environmental outcomes compared with alternative methods.
The Bottom Line: The Technology Helping Save Nature Could Become One of Its Biggest Resource Challenges
Artificial intelligence is creating new opportunities for conservation science.
Organizations can process enormous datasets, monitor wildlife and detect environmental threats more quickly than before.
But these advances are emerging alongside rapid expansion in the data-centre industry.
Reuters found that all 15 environmental organizations it interviewed were using AI while expressing concern about its environmental impacts.
The International Energy Agency’s updated central projection places global data-centre electricity demand near 950 terawatt-hours by 2030.
Water requirements could also increase substantially.
The resulting tension is forcing environmental groups to reconsider how they use technology.
Some are reducing unnecessary computing.
Others are using edge-processing systems or reusing results to limit energy demand.
Greenpeace is questioning the scale of infrastructure expansion itself.
The disagreement does not necessarily mean one side is right and the other is wrong.
It reflects different judgments about the environmental value of particular AI applications.
For conservationists, abandoning useful tools could slow urgently needed work.
For communities affected by massive data centres, accepting unlimited infrastructure growth could create unacceptable local pressures.
And as artificial intelligence helps scientists protect endangered species, monitor forests and recover decades of research, one question is becoming impossible to ignore: Can the world use AI to save the planet without allowing the technology’s growing appetite for electricity, water and land to undermine the very environment it promises to protect?
