The rise of Generative Artificial Intelligence (GenAI) in recent years is reshaping society in many ways. Our education system is no exception. A recent study found that nearly 90 per cent of Irish undergraduate students use AI to assist with assessments, and nearly every fifth student uses AI to generate academic text. This has given rise to new questions about academic integrity, plagiarism and malpractice. Thus, Trinity has, like many educational institutions worldwide, developed a college-wide AI policy, issued in June 2026, regulating how GenAI should be used in the university’s teaching, assessment and research. This policy stipulates, among other things, that AI use is declared in assessments and may be subject to restriction in certain disciplines.
Environmental concerns
However, academic and intellectual originality are not the only issues that a more widespread use of AI in academia entails. The use of GenAI also has significant yet often overlooked environmental impacts. For one, the everyday use of AI — not just the initial training and development of models — consumes significant amounts of fresh water. A US study has found that the cooling of a single AI data centre can consume as much water as a village of 50,000 inhabitants — or one third of the residents in Dublin City Centre, while a 100-word AI prompt can consume up to half a litre of water. While this figure might itself not sound serious, this amounts to a water footprint of over 10,000 litres every day for Trinity’s student body — even with the highly conservative estimate that most students only use AI for a single, short prompt once per day.
This is especially problematic as climate change exacerbates water scarcity, foreshadowed in this summer’s heatwave in the UK and continental Europe, generating protracted droughts. To make matters worse, 25 per cent of the world’s existing and one-third of data centres under construction are located in areas already experiencing water stress. In addition, data centres have climbed to be the 5th biggest energy consumers worldwide this year. This is especially problematic from an environmental point of view, as most of that energy is generated by fossil fuels, adding to global warming and air pollution. At the level of individual user behaviour, a ChatGPT query consumes up to 5 times more energy than a conventional internet search. Thus, while Trinity and Ireland may not feel the immediate impacts of its institutional AI use in the near future, our institutional environmental footprint will have direct repercussions for other people living in already economically, environmentally and socially vulnerable areas.
The marginalisation of sustainability
It is surprising, then, that these environmental considerations feature but at the margins of Trinity’s official statement on the use of GenAI. While vague references are made to the need to reconcile AI use with the university’s Sustainability Strategy 2023–2030, the vast water and energy consumption of AI is mentioned briefly and with a lack of poignant detail at the bottom of the list of potential risks that staff and students should take into consideration when determining how and to what extent to incorporate AI into their teaching and learning, and is notably absent from the advice on how to instruct students to use AI, overshadowed by the emphasis on the risks to privacy, low-quality information, academic integrity, and intellectual development. One may wonder how this statement, coupled with the technology’s aforementioned environmental impacts, relates to Trinity’s Sustainability Strategy, stating that “Our mission is to tackle climate and biodiversity challenges . . . through our educational programmes, our research and innovation, and our day-to-day operations”(emphasis added) and its goals of “advancing our digital estate in support of a dynamic and sustainable academic mission” and being “nature positive” by 2030; one may wonder how the university squares these goals of sustainability, responsibility and ecological preservation with the widespread use of a technology that pollutes environments, warms the planet and guzzles precious fresh water from susceptible communities
A costly policy inconsistency
Now, a college like TCD will understandably emphasise and prioritise academic considerations — such as skill development, labour market requirements and institutional competitiveness — in its AI policy. Thus, the use of AI in education involves an admittedly complex balancing of interests, goals and values. However, considering the urgency and pervasiveness of the climate crisis, sustainability concerns should arguably not be addressed on the sidelines, as if a mere afterthought rather than a pressing and holistic institutional principle. In this way, Trinity’s AI statement seemingly fails to do the complexity of this technology justice.
If the college wants to cohere with its vision of cultivating well-informed and responsible citizens who make a positive contribution to society, it cannot think about the environment last, or overlook its impacts until it hits at home in the future; this very mindset of unsolidarity and short-termism, entrenched in contemporary Western culture and society, is arguably what has caused the climate crisis in the first place. But change is possible, and history has shown that this change often starts in the classroom, with open, self-critical and balanced dialogue.
