AI Could Be Key to Reducing Global Carbon Emissions, New Study Finds
According to a joint study by the London School of Economics and Systemiq, global carbon emissions can be slashed without sacrificing modern lifestyles—and A.I. plays a vital role in the battle to save the planet.
Smart AI applications in the three sectors alone would cut greenhouse gas emissions by 3.2–5.4 billion tonnes annually by 2035, the research found. Contrary to what many people believe, these cuts would be much larger than the emissions generated by AI.
The report, “Green and Intelligent: The Role of AI in Climate Change,” characterises AI not as a tool for incremental efficiency improvements but as a force that can drive the reshaping of the economy to one that values and rewards more sustainable and inclusive considerations.
Net-Zero Is an Opportunity, Not a Cost
Topping off that wish list, the researchers are calling on policymakers and businesses to see the shift to a net-zero economy not as a cost, they say, but as “a remarkable opportunity for innovation, sustainable growth, resilience, and inclusion.”
They looked at three dominant sources of global emission: electricity generation, meat and dairy production and passenger transport. Together, they represent about half of global emissions. The carbon savings possible through the application of AI in just these sectors could offset the 0.4 to 1.6 billion tonnes of annual emissions associated with operating all AI data centers across the globe, we estimate.
As the authors note, “The argument for employing AI to combat climate change is not simply strong — it’s necessary.”
Five Big Ways Artificial Intelligence Is Making a Difference for the Planet
Making Complex Systems Smarter
Energy, transportation and city living all rely on sprawling networks. AI can rethink these systems to function more efficiently.
Consider the brownout whenever wind lags or clouds obstruct the sun’s shine: AI could forecast those differences in renewable energy production and address them in an instantaneous way with demand in sync. For example, DeepMind has already demonstrated that its AI is able to maximise the economic value of wind energy: its technology can boost the value of a wind farm between 13% and 20% through better predictions and forecasts, reducing the amount of backup power required.
Accelerating Discovery and Reducing Waste
Half of the emission reductions required by 2050 won’t come from technologies we have deployed; they’ll come from technologies that are barely out of the lab or blackboard stage, and AI is advancing these breakthroughs.
Look at DeepMind’s GNOME tool, which has already discovered more than 2 million new crystal structures that might transform renewable energy and battery production. Or really, meditate on Amazon’s AI packaging algorithms, which have saved more than 3 million metric tons of material since 2015.
Helping People Make Greener Choices
Our daily choices, whether about eating or how we travel, can together cut emissions by up to 70 percent by 2050. But it’s not always easy to make the right choices.
Breaking down informational silos, AI can serve as a personal environmental coach, offering personalized recommendations. If you’ve ever navigated Google Maps’ eco-friendly options, you’ve already had this benefit. For smart home systems like Nest also employ AI for heating and cooling optimization, which could save millions of tons of CO₂ if scaled up.
Anticipating Climate and Policy Impacts
How should we prepare for a changing climate? AI can crunch huge datasets to predict climate patterns with pinpoint accuracy.
Tools such as IceNet, developed by the British Antarctic Survey and Alan Turing Institute, are using AI to predict sea ice levels in a more precise manner than ever, supporting communities and businesses to get ready. This power also allows for governments to analyze successful international models and develop policies to match.
Protecting Us During Extreme Weather
Early alerts can save lives. As climate disasters become more intense, early warnings are critical. Artificial intelligence systems are being woven into these services and becoming critical safety nets for natural disasters such as floods and wildfires.
Google’s Flood Hub, for instance, relies on machine learning to predict floods up to five days in advance in more than 80 countries—precious time for people to secure and evacuate their homes.
The Numbers Back It Up
The research estimates that, by 2035, AI may:
- Save 1.8 gigatonnes of emissions from the power sector by optimising renewables
- Reduce 0.9 to 3.0 billion tonnes annually through optimization of meat-replicating plant proteins
- Eliminate, through shared mobility and more efficient batteries, up to 0.6 billion tons from vehicles
The Catch: AI Must Be Guided to Develop
But there’s a catch—we can’t just sit back and let market forces shape the evolution of artificial intelligence. The authors suggest an “active state” to make sure that AI is working in the benefit of people and the planet.
They write: “Governments have a central role to play in making sure AI is used effectively to accelerate an equitable and sustainable transition.” That would mean creating incentives for green AI research, regulating its environmental impacts and investing in infrastructure so the benefits can be shared in communities around the world.
Led by innovation and working together across borders, we can unlock AI’s full potential to decarbonize and mitigate the climate crisis—creating a future where both people and the planet can flourish.
