In short
Work out why the first answer missed, then tell the AI what to change in a short follow-up instead of pressing regenerate. Each regeneration is a full new answer, so a habit of rolling the dice five or ten times multiplies the energy of a task, especially for images and video.
Steps
- Read the first answer and name what is wrong with it.
- Reply with a specific correction, such as "shorter", "more formal" or "only the second point".
- Add the missing context or an example if the answer misunderstood the task.
- Stop after two or three attempts and rethink the request, or try a different tool.
Refine, don't restart. Regenerating asks the model to produce a whole new answer from scratch, and it costs the same each time. A short correction usually gets you closer for less.
Why does it matter?
Because repetition multiplies whatever the task costs. For text, each answer is small, but ten attempts are ten answers. For images it adds up fast: in tests by Luccioni, Jernite and Strubell, generating images used on average about 2.9 kWh per 1,000 images, over a thousand times more than classifying text.
How do I get it right sooner?
- Diagnose the miss. Too long? Wrong tone? Missed a fact? Say so: "Keep the structure but cut it to 80 words."
- Give what was missing. A regenerated answer has the same gaps if the prompt does.
- Edit it yourself when only a word or two is wrong.
- For images, change one thing at a time in the prompt rather than spinning for luck.
- Ask for brevity. UNESCO and University College London found shorter prompts and responses could cut energy use by more than half.
Common mistakes
- Treating AI like a slot machine, hoping the next spin is better.
- Asking for "five more options" when you have not said what was wrong with the first ones.
- Carrying on past the third attempt. If it still isn't working, the tool or the request is probably wrong; see whether you need AI at all.
For more, see use AI efficiently.
Sources
- Power Hungry Processing: Watts Driving the Cost of AI Deployment?, Luccioni, Jernite and Strubell, ACM FAccT (2024)
- Smarter, smaller, stronger: resource-efficient generative AI and the future of digital transformation, UNESCO and University College London (2025)
Related
How do I write prompts that use less energy?
How to write AI prompts that need less computation: be specific, set a length, paste only what matters and get it right first time instead of retrying.
How do I cut the impact of AI image and video generation?
Why AI images and video use far more energy than text, and practical ways to generate fewer, shorter, smaller outputs without losing what you need.
Use AI efficiently: everyday habits
Everyday habits that cut the energy of using AI: clear prompts, fewer regeneration loops, care with images and video, caching and turning off defaults.