The environmental ethics of Generative AI: Artificial intelligence or real ignorance?

Blog post

The environmental ethics of Generative AI: Artificial intelligence or real ignorance?

Sarah Clayton, PhD Graduate at University of York Lynda Dunlop, Senior Lecturer in Science Education at University of York Elizabeth Rushton, Head of Education Division at University of Stirling 20 Jun 2025

Artificial intelligence (AI) is increasingly present in debates about education, including at policy level. Earlier this year, the UK Education Secretary (with responsibility for education in England), Bridget Phillipson, set out plans to ‘modernise’ the . At the heart of these plans is a focus on AI, with rhetoric of a ‘great new technological era’ and AI as a ‘radical modernising force for change’ in schools. Such rhetoric is consistent with UK Prime Minister Keir Starmer’s ‘’ to use AI to ‘turbocharge’ a decade of national renewal; but it is perhaps less consistent with the UK commitment to reach net zero greenhouse gas emissions by 2050, given the energy demands of AI. Along with a raft of promised training and guidance for teachers and school leaders, the Department for Education (in England) has updated its , including aspects of intellectual property and product safety. However, the environmental impacts of AI are absent from this guidance and more broadly from debates concerning the use of Generative AI, representing a situation of ignorance. As educators and educational researchers, we argue in this blog post that this absence is of real concern.

Environmental impacts of Gen AI

ChatGPT, Microsoft Copilot and Google Gemini are some of the large language models which are being used in education, for example to . Generative AI has substantial environmental impacts, largely due to the high energy required to train and use large language models, the water that is needed to cool equipment (Gupta et al., 2024), and pollution associated with mineral extraction for the production of chips and servers (Capgemini, 2025). Indeed, Ana Valdivia (2024) uses the concept of ‘supply chain capitalism of AI’ to analyse environmental harms associated with mining, electronics, digital and e-waste associated with AI. This means looking at the environmental impacts of AI at every stage of the supply chain, from mining the minerals used in the electronics hardware, to the emissions associated with the use of the models, to the disposal of hardware when it reaches the end of its life (typically five years).

‘It is estimated that every time ChatGPT is used instead of a search engine query, it uses up to 30 ��times as much energy.’

The energy use associated with AI models comes from multiple sources. Energy is needed to train the models, while energy demand is likely to be significant in the use of these models. To give just one example, it is estimated that every time ChatGPT (prevalent in education settings) is used instead of a search engine query, it uses up to 30 ��times as much energy (Chen, 2025). In a future where large language models are used not just to replace search engines but also for many other applications across different areas, including education, there could be an explosion in ICT-related electricity consumption (Vanderbauwhede, 2023). Increased demand for energy delays the transition away from fossil fuels – but even if our electricity consumption was 100 per cent renewable, use of AI has infrastructure and energy demand that would require more than can be produced by renewables alone. ��Vanderbauwhede (2023) argues for ‘frugal AI’, advocating for responsible and efficient use of AI. This means reducing the use of AI and avoiding AI as a default option.

Implications for educators and education researchers

The UK government frames AI as a transformative force in education in England. Despite the claim that it wants a ‘whole system approach to embedding sustainability through the education system’, consideration of the environmental impacts of AI are absent. This represents a real safeguarding risk as climate change, water scarcity and pollution associated with resource extraction are already causing physical and emotional harms.

For educators and education researchers, AI’s rapid integration demands critical reflection and an understanding of the impacts that using AI will have. As a starting point, we share these reflective questions to help us move beyond real ignorance in the face of the environmental impacts of artificial intelligence:

Who benefits from AI’s expansion in schools? Who doesn’t?
How do we weigh the environmental costs of AI against its potential benefits in the classroom, especially in the context of climate change, water scarcity and environmental pollution?
How should environmental considerations be included in ethical guidelines for the development and use of AI tools in educational and research settings?

References

Capgemini. (2025). Developing sustainable Gen AI. Capgemini Research Institute.

Chen, S. (2025, March 5). How much energy will AI really consume? The good, the bad and the unknown. Nature, 639(8053), 22–24. ��

Gupta, J., Bosch, H., & van Vliet, L. (2024, March 21). AI’s excessive water consumption threatens to drown out its environmental contributions. The Conversation.

Valdivia, A. (2024). The supply chain capitalism of AI: A call to (re) think algorithmic harms and resistance through environmental lens. Information, Communication & Society. Advance online publication.

Vanderbauwhede, W. (2023). The climate cost of the AI revolution.

Sarah Clayton, Mrs

PhD Graduate at University of York

Sarah Clayton trained as a secondary science teacher in 2013. She has taught in Coventry, Salford and Telford and has held a range of middle leadership positions. She completed a master’s degree in education with the University of Manchester and recently commenced a PhD at the University of York, looking at key concepts for understanding climate change in secondary science supported by ��ɫ��’s 2022 doctoral fellowship. Outside of academia, she facilitates workshops around the climate and biodiversity crisis.

Lynda Dunlop, Dr

Senior Lecturer in Science Education at University of York

Lynda Dunlop is Director of Education for Environmental Sustainability at York, and a senior lecturer in science education. Her research is interdisciplinary, working across the social sciences, sciences and arts and humanities. Current studies focus on climate and environmental education, youth political participation in environmental decision making, and public dialogue on climate interventions. She teaches on the MSc Environmental Sustainability Education and Communication. For more information visit https://pure.york.ac.uk/portal/en/persons/lynda-dunlop.

Elizabeth Rushton, Professor

Head of Education Division at University of Stirling

Professor Elizabeth Rushton is a professor in Education and the head of the Division of Education, Faculty of Social Sciences at the University of Stirling. Her research interests are in geography and science education, specifically the professional development of teachers, teacher and student participation in research and policymaking, and education for environmental sustainability (EfES). In 2021, Elizabeth co-led the ��ɫ�� Research Commission with Dr Lynda Dunlop which resulted in the co-creation of a manifesto for EfES with contributions from more than 200 youth, teachers and teacher educators from across the UK. Prior to joining the University of Stirling, Elizabeth was head of Department of Curriculum, Pedagogy and Assessment, Institute of Education, University College London and research director of UCL’s Centre for Climate Change and Sustainability Education. Previously, Elizabeth led the inaugural PGCE secondary geography programme at King’s College London and taught...