In short
Add a short, practical strand rather than a whole new course: explain where AI's energy goes, have students measure the energy of their own code or model runs, and ask them to report compute and energy alongside accuracy. In technical courses, teach efficient methods such as smaller models, quantisation and reuse; in others, focus on sensible use and on judging claims.
Steps
- Add a short session on where AI's energy, water and hardware footprint comes from.
- Ask students to measure energy for at least one coursework run with a tool such as CodeCarbon.
- Require compute and energy to be reported alongside accuracy in technical assignments.
- Teach efficient methods such as small models, quantisation and reusing pre-trained models.
- Include an exercise in checking a published claim about AI's energy use.
A short practical strand in existing courses reaches more students than a new elective. Almost all students now use AI: the Higher Education Policy Institute's 2026 survey found 95% of UK undergraduates said they used it.
What could the strand include?
- Footprint basics. Training versus inference, why images and long inputs cost more, and why published figures differ.
- Measure something. Have students wrap one training or inference run in CodeCarbon and report the result with hardware, run time and location.
- Report like a researcher. Ask for compute and energy next to accuracy, including tuning runs; see report energy in research.
- Efficient methods for technical courses: smaller and task-specific models, quantisation, distillation and reuse of pre-trained models.
- Judging claims for any course: who measured it, per what, and what was left out.
Which frameworks help?
UNESCO's AI competency framework for students includes environmentally sustainable AI. For research groups and departments, the GREENER principles from Lannelongue and colleagues give a framework for governance, estimation and education in sustainable computational science.
What makes the biggest difference
Making energy part of assessment. Once a marking scheme rewards accuracy per kilowatt-hour as well as accuracy, students start to design for it.
Common mistakes
- Treating it as a one-off ethics lecture with no practical task.
- Penalising students who use AI rather than teaching them to use it well.
- Using out-of-date per-query figures.
Read AI literacy in universities and browse courses for material to reuse.
Sources
- Student generative AI survey 2026, Higher Education Policy Institute (HEPI) (2026)
- AI competency framework for students, UNESCO (2024)
- CodeCarbon, CodeCarbon (2026)
- GREENER principles for environmentally sustainable computational science, Lannelongue et al., Nature Computational Science (2023)
Related
AI literacy in universities, including its footprint
Students use AI almost universally while staff training and assessment lag. Why university AI literacy should cover AI's environmental impact.
How do I report the energy use of my AI research?
What to put in a paper about compute and energy: hardware, GPU-hours for all runs, energy, location, carbon, inference cost and the boundary used.
How do I measure the energy AI uses on my own computer or GPU?
Measure local AI energy with hardware counters, GPU telemetry or a plug-in meter, sample over the whole job, and subtract the idle baseline.
How do I learn about green AI?
Where to start learning about green AI, whatever your background: short reads, free tools to try, courses, communities and research to follow.