Our approach
At Cambridge, we are committed to reducing our environmental impact across all areas of our operations, including being carbon zero on all energy-related emissions by 2048.
We recognise that Artificial Intelligence (AI) brings new considerations and implications for the environment. The broad scale and rapid pace of developments in AI also make it more difficult to keep track of its direct and indirect impact on sustainability efforts. It is therefore an ongoing goal for our organisation to understand, monitor and manage the environmental impacts of our AI approaches.
Guidance for leaders and teachers considering the use of AI technologies in school
From our conversations with schools, we are aware that like us, some may also be working through environmental considerations for their organisation.
We do not aim to provide definitive answers here as balancing the pros and cons of AI is nuanced and dependent on the individual situations. However, here is some key information for school leaders and teachers, to help you reflect on the topic of AI and the environment and inform future decisions regarding your use of AI technology in school:
- AI’s impact on the environment.
- Questions for schools to explore when considering sustainable use of AI.
AI’s impact on the environment
As with all new technologies, AI offers both benefits and costs to the environment. Awareness and understanding of the possible impacts is the first step to balanced decision-making when considering sustainable approaches. 'There are high hopes that artificial intelligence (AI) can help tackle some of the world’s biggest environmental emergencies. Among other things, the technology is already being used to map the destructive dredging of sand and chart emissions of methane, a potent greenhouse gas. But when it comes to the environment, there is a negative side to the explosion of AI and its associated infrastructure, according to a growing body of research.' – United Nations Environment Programme
A helpful point to remember is that while AI is a digital technology, it still relies on physical infrastructure. For example, most large-scale AI services are housed in data centres. A data centre is a dedicated space, for example a building, that houses computer systems and their associated components, such as data storage and telecommunications. As demand for AI grows, the number of data centres is also increasing. Physical elements such as these can impact our environment in the following ways:
Energy
Data centres require significant energy input to run. For example, it is estimated that a request made through the generative AI assistant ChatGPT uses 10 times the electricity of a comparable Google search. Due to the large amount of processing power required to train an AI model and run it, the energy used by data centres is significantly increasing. The International Energy Agency 2024 report estimated that electricity use from data centres will increase from 460 terawatt-hours (TWh) in 2022 to more than 1000 TWh in 2026.
Water
The necessary ongoing cooling processes at data centres are highly intensive in their water use. This is concerning given that global water scarcity is a growing problem. Previous research estimated that by 2025, 4.8 billion people would be living with water scarcity.
Materials extraction and waste
While there are significant benefits that computers can provide, it is worth being aware that the quantity of materials required to make computers is significant. One example states that it takes 800 kg of raw materials to make a new 2 kg computer. In addition, once the computer hardware is outdated, there are then concerns with electronic waste, which can contain hazardous substances such as mercury and lead.
Questions for schools to explore when considering sustainable use of AI
The following questions may help you consider and balance the need for different AI technologies in light of environmental considerations:
Consider the context
The need:
- Is there a genuine need to use AI or are there alternatives?
The costs:
- What are the potential social (including educational), economic and environmental costs of adopting new AI technology?
The benefits:
- What are the potential social (including educational), economic and environmental benefits of adopting new AI technology?
Computing infrastructure and technology use
Think about the following.
Infrastructure:
- Have you got sufficient computing infrastructure and hardware already in place or do you need new equipment?
- What is your plan for any redundant equipment?
- How might you help reduce the energy and water demands of any computing infrastructure or equipment you choose?
Use:
- Have you got a clear purpose and limited scope for the AI use, so you can use the minimum viable computing model required and reduce the occurrence of wasted functionality? For example, why invest in a tool that can process images if you only need written outputs?
- Have you explored the most energy-efficient processes for your AI use, such as the option to use it offline? Is there an option to work with pre-trained models?
Support your people
Training:
- Have you got capacity to train staff and students on using AI tools efficiently and effectively? For example do the people in your school know how to write effective prompts?
Policy:
- Do you have a policy or guidance regarding appropriate AI use for your school, staff and students, including environmental considerations?
Join our Climate change education research community
Would you like to continue the conversation around AI and the environment? Join our research panel and share ideas with fellow school leaders and teachers.