Sustainability Concerns

AI has the potential to support sustainable development goals, but it also comes with significant sustainability concerns. As AI tools become more embedded in the research process, we must ask: how do we harness AI for sustainability without deepening the harm it causes?

AI systems, especially large language models like ChatGPT, require enormous amounts of energy and water to train and run. Do you know that one ChatGPT query uses roughly five times more electrical power than a simple internet search?

Most of AI’s resource demand comes from:

  • More data centres that power AI. They use mostly fossil-fuel electricity and consume vast amounts of water for cooling.

Global electricity demand from data centres is set to more than double over the next five years, consuming as much electricity by 2030 as the whole of Japan does today.

Fatih Birol IEA Executive Director

Carbon emissions from AI training are steadily increasing. Training early AI models, such as AlexNet (2012), had modest amounts of carbon emissions at 0.01 tons. More recent models have significantly higher emissions for training: GPT-3 (2020) at 588 tons, GPT-4 (2023) at 5,184 tons, and Llama 3.1 405B (2024) at 8,930 tons. For perspective, the average American emits 18 tons of carbon per year

(Stanford AI Index Report 2025, p. 13)
  • Hardware production: The computers that run AI need rare minerals like gold, tantalum and tungsten, which are often mined in environmentally destructive ways.

Research also reveals that AI's environmental impact extends beyond data centers to encompass a complex global supply chain that extracts resources, manufactures components, and exploits labor across multiple industries and countries. Her fieldwork in Querétaro, Mexico illustrates how AI data centers consume local water and energy supplies, creating scarcity for surrounding communities like Maconi. This situation highlights the often hidden environmental justice concerns that arise when technological advancement conflicts with basic resource needs of local populations.

 

Can AI Still Be Part of the Solution?

AI can do in seconds what might take a team of experts a year. This is why we must harness it to reverse the damage we’ve done to the planet.

Professor Anil Madhavapeddy (ai@cam)

Despite its costs, AI also holds real promise for sustainability:

  • It’s enhancing climate and weather forecasting while consuming less energy than traditional models.
  • It helps monitor deforestation, ocean health, and natural disasters at a scale that would be unmanageable manually.
  • While AI data centers will increase electricity demand and emissions, this increase will be relatively small within the overall energy sector and could potentially be offset by AI-enabled emissions reductions if widely adopted. (IEA’s Energy and AI Report 2025)
  • AI is becoming increasingly integral to scientific discovery and could accelerate innovation specifically in energy technologies such as batteries and solar PV. (IEA’s Energy and AI Report 2025)
  • AI hardware's 40% annual energy efficiency improvement offers a promising path to more sustainable computing (Stanford AI Index Report 2025)
  • Universities and tech companies are also exploring greener practices—like reusing server heat, reducing water use, and developing more efficient algorithms.

AI isn’t just a digital tool—it’s a physical, global system. If we want to use it responsibly, we need to treat sustainability as a core part of how AI is designed, deployed, and governed. For behavioural researchers, AI offers powerful capabilities while raising important sustainability considerations. Rather than refraining from using these tools, implement mindful practices like batching queries, or selecting appropriately-sized and more energy efficient models for your needs.

 

This article was published on