Will AI do More Harm than Good to the Environment?
Research from the London School of Economics and Political Science (LSE) and Systemiq shows AI can play a powerful role in the climate transition.
The study specifically focuses on the power, transport and food sectors, making up nearly half of global emissions.
By 2035, it found that AI could reduce global emissions by 5.4 GtCO₂e annually, outweighing AI’s own energy use.
”From enabling smarter logistics to optimising energy grids, AI is already driving efficiency,” said Sophie Graham, Chief Sustainability Officer at IFS, on LinkedIn.
“But its impact goes further, equipping us with powerful advances in forecasting and early detection for severe weather events, critical to long-term resilience.
“IFS Industrial AI has a clear role to play in this transition, supporting asset-intensive industries adapting to a low-carbon future.
“Now is a pivotal moment to scale AI responsibly and equitably - especially where it can deliver the greatest climate value.”
Why AI matters in the climate transition
AI, the research says, could be a powerful enabler in the transformation to sustainability due to its broad applicability and ability to scale innovation rapidly.
It could not just work as a technological tool, but drive systemic transformation.
This could help to boost economic growth, break cycles of under-investment and achieve emissions reductions.
The study used a bottom-up approach to estimate AI’s emissions reduction and impact in power, food (meat and dairy) and mobility (light road vehicles).
In the power sector, AI-enabled grid management and efficiency improvements could abate up to 1.8 GtCO₂e annually by 2035.
In meat and dairy, support from AI to adopt alternative proteins could support emissions savings from 0.9-3.0 GtCO₂e annually.
Light road vehicles could be supported by AI-powered shared mobility and advances in EV adoption, potentially cutting emissions by up to 0.6 GtCO₂e annually.
This is a total of up to 5.4 GtCO₂e reductions annually.
AI in the power sector
The power sector has a high share of global greenhouse gas emissions and integrating renewable sources is crucial to achieving net zero according to the IEA.
Intermittent renewables, like solar and wind, require careful real-time balancing with demand.
AI can help to forecast electricity supply from renewables and match it to demand alongside managing distributed energy resources like EVs and batteries.
The study cites that Google DeepMind’s AI increased wind energy’s ergonomic value by 20% through reducing the need for standby backup power.
Optimising operations can also help to increase the load factor of wind and solar power plants, the study says.
This means more clean energy is generated from the same assets with lower emissions per unit of energy generated.
The environmental impacts of AI
Despite this decarbonisation potential, AI has an environmental impact of its own.
Data centres are the backbone of AI, where models are trained and run.
These tasks can require significant amounts of energy both to power the data centres and cool them.
As AI adoption expands, so does the demand for data centre capacity and, consequently, electricity.
LSE and Systemiq’s study, however, shows that its potential reductions outweigh the estimated extra emissions associated with increased data centre power use for AI applications.
This is estimated to fall between 0.4-1.6 GtCO₂e and potential reductions could fall between 3.3-5.4 GtCO₂e.
While the research shows AI’s emission reduction potential is large, the analysis only covers three sectors and does not model system-wide spillover or rebound effects.
It also does not account for AI’s impacts on investment or job creation.
“Policymakers must create enabling conditions for AI deployment, provide financial incentives for research and development, and ensure that AI applications are directed toward public goods and high-impact areas,” the paper says.
