AI's Environmental Impact Is Huge. What Can We Do About It?

For most Australians, artificial intelligence (AI) is already part of our lives. It’s embedded in maps, cameras, banking apps, customer service chats and more.

For businesses, government organisations and charities like Greenfleet, AI is increasingly woven into everyday operations. While the responsible use of AI presents huge opportunities for productivity, the environmental impacts can’t be ignored.

That’s why, as an environmental not-for-profit, we are thinking deeply about how we use AI and how we help our supporters compensate for its impact.

The Negative Impacts Of AI Cannot Be Ignored

As the use of AI-powered technology accelerates, so does its environmental footprint. These new tools rely on energy-intensive data centres that generate significant carbon emissions and place pressure on water systems and supply.

Every time you enter a prompt into a large language model (LLM), it produces ‘tokens’ that use computational resources and contribute to the data centre’s energy and water use. The number of tokens produced depends on the length of your prompts and the size of the output. This means, how you use AI determines your impact.

As published in The Guardian, Dutch researchers estimated that AI use accounted for 80 million tonnes of carbon emissions and a potential 765 billion litres of water use in 2025, demonstrating the scale of its environmental impact.

In addition to this environmental crisis, societal challenges are also emerging with changes to the job market and potential effects on our critical thinking capabilities. It’s vital to acknowledge these realities, while also understanding the opportunities that come with the responsible use of AI.

How Are Industries Using AI?

AI has huge potential for good in many industries. In agriculture, precision farming gathers data from drones, sensors and satellites to monitor soil, weather and crop conditions, which can help farmers to apply resources like water or fertiliser specifically when and where they are needed.

In healthcare, AI is being used for cancer screening to detect cervical, breast and prostate cancer. General practitioners are using artificial scribes and receiving support with information about diseases, treatment pathways, diagnoses and tests.

Elsewhere, AI can reduce emissions from storage, transport and waste generated from global supply chains by helping companies forecast demand more accurately, optimise container usage and reduce excess inventory. AI tools are also being applied to product design so manufacturers can refine products for better durability and resource use.

Increasingly, organisations are considering how AI might benefit their work. Greenfleet supporter Whisker Architecture has been thinking deeply about their environmental footprint and is conscious of AI’s impacts.

“AI is rapidly becoming part of professional practice across almost every industry, including architecture,” said Director and Registered Architect Audrey Whisker.

“At Whisker Architecture, we use new tools thoughtfully where they improve efficiency and allow more time to focus on human-centred design and the ongoing development of our business. We already offset part of our environmental impact through Greenfleet and believe businesses should thoughtfully consider the broader impacts of the tools they choose to adopt.”

The technological advancement of AI has potential to positively change how we live and work, particularly when its use is considered and responsible. For environmental organisations, the opportunities are broad.

AI Opportunities For Environmental Organisations

While the environmental effects are undeniable, AI presents opportunities for organisations like Greenfleet that are fighting climate change and habitat destruction.

Our work involves planting forests in Australia and New Zealand that we legally protect and monitor over time. AI can increase accessibility and enhance our monitoring efforts by processing satellite, acoustic and image data to monitor ecosystems. This analysis can help with decision making as we better understand what’s happening at our projects.

AI-enhanced patrol routing, surveillance and alert systems are already being used in other parts of the world to reduce poaching and improve response times for critically threatened species. Parks Victoria is utilising a new tool to rapidly analyse remote camera images to identify and protect native wildlife.

Technology like this can deepen our understanding of the habitat we’re creating for endangered species, while offering additional collaboration opportunities with ecologists and researchers who use our forests as ‘living laboratories’.

Greenfleet forests are legally protected for 100 years and AI technologies can enhance the way we monitor our forests and the wildlife inhabiting them, so they thrive well into the future.

Managing the Environmental Impact of AI

With the opportunities and benefits that AI offers, come the unavoidable environmental impact of increased AI use. Ignoring AI altogether will not stop its expansion, nor will it protect our rivers, forests and climate, which is why Greenfleet is committed to taking action. But what does action look like?

From individuals to sustainability advisors and executive leadership teams, figuring out how to reduce the environmental footprint of AI can be daunting. The pressure to utilise AI effectively is increasing and the tech companies behind these tools provide little transparency on the impact of their products.

At Greenfleet, we’ve been thinking deeply about how we approach this tidal wave of new technology; both as an organisation which can benefit from it, and as a not-for-profit whose purpose is to remove carbon and restore nature.

First Ask: Do I Need To Use AI For This?

The first step in managing your environmental footprint is always to reduce emissions wherever possible.

The impact associated with AI is significantly higher than other technologies such as web search, email, or social media. In 2024, Goldman Sachs estimated that a query to an LLM consumes roughly 10 times more energy as the same search on Google.

We need to keep this in mind when considering whether to use an AI tool for daily tasks or continue using other, less emission-intensive tools. Just as organisations choose when to fly staff to an event and when to have them connect virtually, we can make conscious decisions about when we do and don’t use AI.

Using AI Responsibly 

Technology is evolving quickly so it can be hard to know the right time to put thought into AI adoption.

The good news is, responsible AI frameworks are already being developed to help organisations manage risk, mitigate negative environmental and social impacts, and focus on transparency, accountability and human-centric design.

Law Squared, a law firm working with organisations across the lifecycle of AI adoption, told us what they’re hearing from organisations thinking about responsible AI use.

"We're seeing organisations move from asking whether to use AI to asking how to use it responsibly - and that's the right question," said Vladimir Kravchenko, Commercial Lawyer and AI Advisory Practice Lead at Law Squared. "The Australian Government's AI Ethics Principles and the Voluntary AI Safety Standard point in the same direction: transparency, accountability, human oversight, and a clear-eyed view of the impacts AI creates beyond the screen."

As a purpose-led, people-first charity, it’s vital that Greenfleet’s AI use is responsible. We used the Australian Government’s set of AI ethics principles to inform our approach to the adoption of this technology.

At Greenfleet, we use AI to support research, analyse datasets and reduce time spent on administrative tasks to free our team up to focus on more impactful work. We are actively engaging with our team to assess which tools are most valuable and auditing the effectiveness of those currently in use.

We want to make sure our people have the tools they need, while staying true to our purpose.

Thoughtful AI Use Can Reduce Your Impact

There are simple ways we can all minimise AI-related environmental impact. For example, putting care into how we prompt can reduce token creation and lower our environmental footprint.

Some advice we’ve taken on at Greenfleet:

  • Batch process tasks to reduce query amounts: Include your entire request in one prompt, instead of multiple. For example, “Find the five most recent journal articles on AI data centres and summarise their key findings”.   
  • Be specific and goal-oriented: Start with a ‘verb’ and use numbers and key words to clearly outline what output you want. For example, “List three key impacts of climate change on agriculture, each under 20 words.”
  • Preference estimates when precision is not required: Ask for approximates as precision requests demand higher querying power. 
  • Use AI models that run on your device: third-party cloud servers can rely on more energy-intensive, remote data transfers.
  • Consider the AI tools that you are using: investigate whether the provider publishes data on their energy, carbon and water usage, and whether they utilise renewable energy to power their data centres.

Increasingly, businesses are using agentic AI that automates defined, repetitive tasks, eliminating the need for prompts or other manual intervention. Tools like this require even more thoughtful use as there are fewer opportunities for emissions-saving intervention. Thankfully, there are still other ways to compensate for the environmental impact.

When Using AI, Find Trusted Ways To Offset The Impact 

Ultimately, with AI becoming an almost inescapable part of our lives, we must offset those emissions we cannot avoid, while also improving our water systems.

That’s why, in an Australian-first, we have developed AI ImpactCover to enable authentic climate action. It offsets the emissions generated through your AI use through native reforestation and includes an additional contribution to support water quality.

Greenfleet is Australia’s first carbon offset provider with a near 30-year track record of removing carbon emissions by planting native, biodiverse forests on behalf of our supporters. With legal protection for up to 100 years, these forests restore habitat for wildlife and help fight the impacts of climate change.

With AI ImpactCover you can trust your emissions are being offset through tangible and authentic climate action.

Discover AI ImpactCover

How To Offset AI Emissions

For AI ImpactCover, Greenfleet partnered with independent sustainability consultants Cool Planet, to develop a model to offset the environmental impact of AI use.

Cool Planet specialises in carbon accounting and impact modelling. Their calculations include five emissions sources: server energy, data centre overhead, hardware, data centre location, and the device the query is made on.

The model takes into account that most Australian AI queries currently go to data centres in the US and can be updated over time to reflect the changing nature of energy usage and intensity for Australian consumers.

"Quantifying AI's carbon and water footprint required careful, conservative methodology," said Dan Harper, Founder and CEO, Cool Planet. "This isn't a space where approximations are good enough."

Cool Planet’s model considered data published by the AI-providers and is based on peer-reviewed research. We’ve adopted deliberately conservative assumptions to establish a robust and defensible estimate of the emissions associated with average annual AI use. 

"AI ImpactCover is a credible, well-structured response to a genuine and growing problem, and Cool Planet is proud to have contributed the emissions modelling that underpins it," said Dan.

Environmental Cover For Your AI Use: AI ImpactCover 

AI ImpactCover is a comprehensive solution to help individuals and businesses use AI in an environmentally conscious way. Whether you’re just starting to use AI tools, or already rely on them substantially, it allows you to compensate for the impact you’re making.

"What Greenfleet is doing with AI ImpactCover gets at something the frameworks don't yet fully address,” said Law Squared’s Vladimir Kravchenko. “Responsible use means taking ownership of the environmental footprint of the tools we choose, not just the outputs they produce.”

In partnership with Cool Planet, we established a two-tiered approach for AI ImpactCover: one option for light to moderate AI usage, for those who occasionally use AI for research or productivity, and another for heavy usage, ideal for organisations using agentic AI, for example. Each tier directly supports the restoration of native Australian forests to remove your emissions with an additional contribution to plant riparian trees that support water quality enhancement.

It’s not possible to offset water use in the same way we can carbon emissions, but we’ve included a contribution to support riparian restoration and water quality enhancements within Greenfleet projects. This will occur at the Strzelecki Nature Link on Boonwurrung Country in Victoria.

With AI ImpactCover you can ensure that when you’re using AI today, you’re growing climate hope for tomorrow.

Discover AI ImpactCover  

Use AI Responsibly While Growing Climate Hope

As an environmental not-for-profit, we are thinking deeply about these issues because we must. The landscapes we work in and forests we restore exist in the real world, not the cloud. AI can help solve problems – but only if we are honest about the ones it creates. 

Consider AI carefully, use it responsibly, and remember to offset the emissions you can’t avoid. Together, we can grow climate hope.