In short
Ask whether a simpler tool already does the job reliably: a search engine, calculator, map, dictionary, template or spreadsheet formula. Use generative AI where it saves real time or improves the result, such as drafting, explaining or reworking text, and skip it for look-ups and sums that other tools handle well.
Steps
- Write down in a sentence what you need the result to be.
- Check whether a search, calculator, map, dictionary or template already gives it.
- Use AI when the task needs drafting, explaining, summarising or reworking, and the result can be checked.
- Avoid AI when you need an exact, checkable fact or figure that a trusted source already provides.
- Review tasks you repeat often, because the choice matters most for them.
The cheapest AI request is the one you don't need to make. Generative AI is useful, but for many everyday tasks a simpler tool is quicker, more reliable and uses less energy. A 2026 UK green AI sandpit of mathematicians, computer scientists and industry kept returning to the same point: people often reach for AI when a well-understood method already exists.
What is a quick test?
| If you need… | Usually better | AI earns its place when… |
|---|---|---|
| A fact, time, address or price | Search or the official site | You need it explained in plain words |
| A sum or conversion | Calculator or spreadsheet | The maths is buried in messy text |
| Directions | A map app | Never, really |
| A standard letter | A saved template | Each letter needs real tailoring |
| A first draft, summary or rewrite | AI | Usually: this is what it is good at |
Why does it matter?
General AI models do a lot of work for every request. Luccioni, Jernite and Strubell found that multi-purpose generative models were orders of magnitude more expensive to run than task-specific systems for the same task. And AI is not free even when it helps: a 2025 study of programming tasks found AI-assisted coding averaged about 33 times the carbon footprint of the manual approach in the cases tested.
That does not mean never use it. It means use it where it pays back. The W3C Web Sustainability Guidelines make the same point for people building products: evaluate the need for AI before implementing it.
Common mistakes
- Using a chatbot as a search engine for simple facts, then having to check them anyway.
- Asking AI to do arithmetic a calculator does exactly.
- Defaulting to AI out of habit for tasks you repeat daily, where a template would do.
For the bigger picture, read efficiency is not sufficiency and choose the right model.
Sources
- Web Sustainability Guidelines, W3C (2025)
- Power Hungry Processing: Watts Driving the Cost of AI Deployment?, Luccioni, Jernite and Strubell, ACM FAccT (2024)
- Comparative analysis of carbon footprint in manual vs. LLM-assisted code development, Cheung, Kaul, Jahangirova, Mousavi and Zie, ResponsibleSE 2025 (ACM) (2025)
Related
How do I reduce the energy my ChatGPT use consumes?
Practical ways to cut the energy of using ChatGPT and similar chatbots: shorter exchanges, fewer regenerations, lighter models and less image generation.
How do I choose a smaller AI model?
How to pick the smallest AI model that does your task well: set a quality bar, test small to large, check energy ratings and keep a big model as fallback.
Choose the right model for the job
How to right-size AI: when to use small or specialised models, when a frontier model earns its cost, how routing works, and when not to use AI.
Efficiency is not sufficiency: the rebound effect in AI
Why making AI more efficient does not guarantee it uses less energy overall, what the rebound or Jevons effect is, and what helps keep savings real.