greenai.guide

Home › Articles › AI literacy in universities, including its footprint

Article

AI literacy in universities, including its footprint

Updated · 4 min read

In short

Nearly all students now use generative AI, but universities' strategies, staff training and assessment have not kept pace. AI literacy, knowing how AI works, when to use it and what it costs, is becoming an expectation and in the EU a legal duty. Including AI's environmental impact in that literacy helps staff and students use it well, not just more.

Key points

  • In a 2026 UK survey, 95% of undergraduates said they used AI and 94% had used generative AI for assessed work.
  • The EU AI Act includes an AI literacy duty for organisations that provide or use AI, applying since February 2025.
  • UNESCO's AI competency frameworks for students and teachers both include environmentally sustainable AI.
  • Some universities have made AI training mandatory for all staff; environmental impact is rarely part of it yet.
  • Assessment that can be completed by AI needs redesign, and the most robust options cost staff time.

Universities are where many people first learn to use AI seriously, and where the next generation of AI developers is trained. They are also large organisations with thousands of staff using AI tools every day. That gives them a double role in green AI: teaching people to use AI well, and using it well themselves.

How widely do students use AI?

Almost universally. The UK's Higher Education Policy Institute (HEPI) surveys undergraduates each year:

2025 survey 2026 survey
Students using AI 92% 95%
Students who have used generative AI for assessed work 88% 94%

In the 2025 survey, 42% of students said staff were well equipped to help them work with AI, up from 18% the year before. In the 2026 survey, 38% said their university provided AI tools.

The pattern is similar elsewhere. In the United States, EDUCAUSE's 2025 AI landscape study found that 57% of institutions saw AI as a strategic priority, up from 49% the previous year, while very few had found new funding for it.

Do universities have an AI strategy?

Many are still working one out. The UK's Jisc published a strategic framework for AI in colleges and universities in 2025, explicitly to help institutions avoid piecemeal approaches. A 2026 UK green AI sandpit of mathematicians, computer scientists and industry heard the same picture: good practice exists, but scattered. Some institutions check every assessment for exposure to AI, some provide AI literacy for students, others for staff, and few do all of it.

What is AI literacy, and is it required?

AI literacy is knowing enough about AI to use it well: how it broadly works, what it is good and bad at, its risks, how to check its output, and what it costs. In some places it is becoming a duty.

  • European Union. Article 4 of the AI Act, which has applied since 2 February 2025, places an AI literacy duty on organisations that provide or deploy AI systems, which includes universities using AI tools. The Commission's guidance says there are no strict requirements or mandatory training formats. The article's wording was amended in 2026 to require organisations to support AI literacy, so check the current text.
  • United Kingdom. There is no legal duty, but the Russell Group's 2023 principles, backed by its 24 vice-chancellors, commit universities to support students and staff to become AI-literate.
  • International. UNESCO published AI competency frameworks for students and for teachers in September 2024.

Why include environmental impact?

Because how people use AI affects how much energy it uses, and most users have no idea. Users choose models, write prompts and decide how many times to regenerate an answer. Each choice changes the energy cost. See use AI efficiently.

UNESCO's frameworks already include it: the student framework has a principle of encouraging environmentally sustainable AI, and the teacher framework one of promoting trustworthy and environmentally sustainable AI.

Environmental content also helps with credibility. Some students and staff are sceptical of generative AI on ethical or environmental grounds. Training that takes those concerns seriously, with evidence, is more persuasive than training that ignores them. Useful topics include:

Should staff training be compulsory?

A growing number of universities think so. The University of Leeds, for example, launched mandatory generative AI training for all staff and students in September 2025.

The comparison often made is with data-protection training: not everyone needs to be an expert, but everyone needs to know enough to act responsibly. The sandpit discussion drew the same comparison with GDPR training. Public examples of compulsory training that covers environmental impact as a topic, rather than as a concern raised in feedback, are still hard to find. That is a gap worth filling.

What does AI mean for assessment?

Any assessment a student can complete with AI, unsupervised, no longer shows reliably what the student can do. With almost all students using AI for assessed work, the question is not whether to respond but how.

Common responses:

  • Redesign: tasks that use AI openly and assess judgement, critique and process.
  • Secure assessment: in-person exams, vivas and practical tests.
  • Dialogue: one-to-one conversations about the work, which are robust but cost staff time.

The most robust options need more staff time. One argument heard at the sandpit is that time saved by using AI well elsewhere in teaching and administration could pay for it. That depends on staff being trained to use AI efficiently, which brings the argument back to AI literacy.

What about teaching green AI itself?

Computing, data science and mathematics courses can teach efficient AI as a technical skill. Topics such as measuring energy, model compression, efficient training and carbon-aware scheduling fit naturally into existing modules. Several partner universities in GreenCode, a European research project on energy-efficient software, teach courses on energy-efficient and sustainable software. The courses directory lists courses on green AI and sustainable computing, and the PhDs directory lists research opportunities.

What to do next

  • Include AI's environmental impact in any AI literacy training for staff and students.
  • Make staff training cover how to use AI efficiently: model choice, prompting, avoiding repeated generation.
  • Review assessments for exposure to AI, and budget the staff time robust alternatives need.
  • If you are in the EU, check your AI literacy measures against Article 4 of the AI Act.
  • Teaching younger students? See the For schools section and lesson ideas.
  • Researchers: see the guide for researchers.

Frequently asked questions

What is AI literacy?

The knowledge and skills to use AI well: understanding broadly how it works, what it is good and bad at, its risks, how to check its output, and its costs, including its environmental footprint.

Do universities have to provide AI literacy training?

In the EU, the AI Act requires organisations that provide or deploy AI systems to take measures on AI literacy for their staff, and universities are deployers. Elsewhere there is usually no legal duty, but sector guidance, such as the UK Russell Group's principles, expects it.

Should AI literacy include environmental impact?

Yes. UNESCO's competency frameworks include environmentally sustainable AI, and understanding that model choice, prompt length and repeated generation affect energy use helps people use AI more efficiently.

Sources

  1. Student generative AI survey 2026, Higher Education Policy Institute (HEPI) (2026)
  2. Student generative AI survey 2025, Higher Education Policy Institute (HEPI) (2025)
  3. AI literacy: questions and answers, European Commission (2025)
  4. AI competency framework for students, UNESCO (2024)
  5. Russell Group principles on the use of generative AI tools in education, Russell Group (2023)
  6. A strategic framework for AI in colleges and universities, Jisc National Centre for AI (2025)
  7. Embracing AI at Leeds: a shared journey for our community, University of Leeds (2026)
  8. 2025 EDUCAUSE AI landscape study, EDUCAUSE (2025)
  9. Project results, GreenCode (2026)

Join us

Help cut AI's impact on the planet

Get new guides, funding calls and directory listings by email, about once a month. Or add your product, course, PhD or research to the directories.

No tracking. Unsubscribe in one click. Other ways to join · Privacy