# Green AI for pupils

> A short, friendly guide for young people: what green AI is, why AI uses energy and water, smart habits, and how to find out more.

Source: https://greenai.guide/schools/for-pupils/  
Updated: 2026-11-19

## In short

Green AI means using artificial intelligence in a way that wastes as little energy, water and equipment as possible, and only using it when it really helps. Every time you ask an AI chatbot something, computers in a big building far away do the work, and that uses electricity. You can make a difference with a few smart habits.

## Key points

- AI chatbots run on computers in buildings called data centres, not just on your phone.
- One question uses a tiny bit of energy, but billions of questions every day add up to a lot.
- Making pictures and videos with AI uses much more energy than short text answers.
- Smart habits: ask clearly once, ask for short answers, and choose a simpler tool when it will do.

## What is green AI?

Green AI is about using AI without wasting energy, water or equipment, and only using it when it really helps.
It is not about never using AI. It is about using it well.

AI means computer programs that can do things that seem clever, like answering questions, recognising faces or
making pictures. The chatbots you may have heard of are one kind, called [generative AI](/glossary/#generative-ai),
because they generate (make) new text, pictures or sounds.

## Why does AI use energy?

Because the thinking happens on powerful computers, and computers need electricity. When you type a question into
a chatbot, your phone or laptop sends it a long way to a building called a [data centre](/glossary/#data-centre).

A data centre is a huge building packed with computers. They get hot, so they need cooling, which uses more
electricity and often water. Then the answer travels back to you.

How much energy is one question? Not much. Google says a typical text question to its Gemini chatbot uses 0.24
watt-hours. OpenAI says a typical ChatGPT question uses about 0.34 watt-hours, like an energy-efficient light
bulb left on for a couple of minutes. But billions of questions are asked
every day, and all the data centres in the world together use about 1.5% of the world's electricity, according to
the International Energy Agency. That share is growing fast.

## Does AI use water too?

Yes, a little for each question, and a lot in total. Many data centres use water to keep their computers cool.
Google says a typical Gemini text question uses about 0.26 millilitres, roughly five drops. That sounds tiny,
but some data centres are built in places that are already short of water, so where they go matters.

"The cloud" sounds light and fluffy, but it is really buildings, cables and computers on the ground, using real
electricity, water and land.

## Why do some AI tasks use more energy than others?

Bigger jobs need more computing. AI reads and writes in small chunks of text called [tokens](/glossary/#token).
A token is often part of a word, about four letters long in English. A long answer has more tokens, so it takes
more work.

Pictures and videos take much more work than words. In one study, the least efficient picture-making AI tested
used about half a phone charge of energy for every picture it made.

## What smart habits can I use?

Most habits that save energy also get you better answers. Here are six to try.

1. **Ask yourself: do I need AI for this?** For a quick fact, a search or a book might be faster. For a sum, use a
   calculator.
2. **Ask clearly, once.** Say exactly what you want. One good question beats five vague ones.
3. **Ask for short answers** when short is all you need. A study for UNESCO found that shorter questions and
   answers can cut energy use by more than half.
4. **Think before making pictures or videos.** Do you really need 20 versions?
5. **Don't use AI just to pass the time.** Chatting to a chatbot for fun all evening adds up.
6. **Check the answer.** AI can be wrong. If you have to ask again and again, that wastes energy and time.

And always follow your school's rules about using AI, and the age limits the apps themselves set.

## Can AI help the planet too?

Yes. AI is used to forecast weather, spot leaks in water pipes, plan where to put wind farms and study wildlife.
The International Energy Agency says AI could help cut emissions in many industries, but it also warns that AI
is "not a silver bullet".

So the big question is not "is AI good or bad?" It is "are we using AI for the right things, in the right way?"
That is what green AI is about.

## How can I find out more?

There is lots more to explore, and people are needed to solve these problems.

- Look up words you don't know in the [glossary](/glossary/).
- Read our page for [everyone](/guides/public/) about using AI wisely at home.
- Ask your computing, science or geography teacher about it. They may try one of our
  [lesson ideas](/schools/lesson-ideas/).
- If you like computers, maths, science or the environment, you could study green computing later. See the
  [courses list](/courses/) for what universities offer.

## What to do next

- Try one smart habit from the list above this week.
- Next time you use AI, ask yourself: is this the right tool?
- Tell a friend or family member one thing you learned about how AI uses energy.
- Look up three new words in the [glossary](/glossary/).

## Frequently asked questions

### Does using AI hurt the planet?

It has a cost, because the computers that run AI use electricity, water and materials. One question is a small cost, but lots of people asking lots of questions adds up. Using AI wisely keeps that cost down.

### Is it bad to use AI for homework?

Follow your school's rules first. If AI is allowed, use it to help you learn, not to do the work for you, and think about whether a book, a website or your teacher would be better.

### What job could I do in green AI?

People work on this as computer scientists, mathematicians, engineers, environmental scientists and policy experts. Studying computing, maths, science or geography is a good start.

## Sources

- [Measuring the environmental impact of delivering AI at Google scale](https://arxiv.org/abs/2508.15734), Google (arXiv:2508.15734) (2025)
- [The Gentle Singularity](https://blog.samaltman.com/the-gentle-singularity), Sam Altman (OpenAI) (2025)
- [Energy and AI](https://www.iea.org/reports/energy-and-ai), International Energy Agency (2025)
- [Power hungry processing: watts driving the cost of AI deployment?](https://arxiv.org/abs/2311.16863), Luccioni, Jernite and Strubell (arXiv:2311.16863) (2024)
- [What are tokens and how to count them?](https://help.openai.com/en/articles/4936856-what-are-tokens-and-how-to-count-them), OpenAI Help Center (2025)
- [AI large language models: new report shows small changes can reduce energy use by 90%](https://www.unesco.org/en/articles/ai-large-language-models-new-report-shows-small-changes-can-reduce-energy-use-90), UNESCO (2025)
