Feel Guilty Using AI? Here’s the Real Environmental Impact, and How to Shrink It

If you care about the climate, using AI can feel uncomfortable. You’ve seen the headlines about data centers, power grids and water. And now your work, school or friends expect you to use it anyway.

That guilt comes from a good place. But it helps to know what actually matters, so you can put your energy where it counts.

The honest numbers

One AI text answer is small:

• Google measured 0.24 watt-hours and 0.26 mL of water for a median Gemini text prompt, about nine seconds of TV and five drops of water.
• OpenAI says an average ChatGPT query uses about 0.34 watt-hours.

The problem is scale and growth. The International Energy Agency estimates data centers used about 485 TWh of electricity in 2025 and expects roughly 950 TWh by 2030. Where that electricity comes from decides most of the climate impact.

So your individual question isn’t the crisis. The way AI is built and powered at scale is. That’s good news: it means your choices about how and where you use AI matter more than whether you use it at all.

What makes the biggest difference

1. What powers the data center. The same answer on hydro or wind has a fraction of the carbon of one on coal or gas. This is the biggest factor.
2. What you ask for. Images, video, long documents and “deep thinking” modes use many times more energy than a short text answer.
3. How big the model is. Small models use much less energy per answer.
4. How you use it. Regenerating the same answer ten times costs ten times as much.

7 ways to shrink your AI footprint

1. Use a small, fast model for simple questions. Save the biggest models for hard problems.
2. Skip image and video generation unless you need it.
3. Ask clearly once. A good prompt beats five regenerations.
4. Send only what’s needed. Paste the relevant page, not the whole 200-page report.
5. Turn off “reasoning” modes for everyday questions.
6. Use on-device AI when you can. Apple Intelligence on recent iPhones and Macs runs with no data center at all. See how to chat with Apple Intelligence.
7. Choose a provider that runs on clean energy and tells you where.

Choose an AI that’s working on the problem

Viro AI was built for people who feel exactly this tension:

• Viro’s own model runs in data centers in Iceland, Norway, Finland and Ireland, powered by hydro, geothermal and wind, with cooling that relies on cold air and closed loops rather than evaporating water.
• ChatGPT, Claude and Gemini are included too; for those, Viro funds new clean energy to match the electricity each message uses and labels them as energy matched.
• Every reply shows its estimated energy, so you can see your footprint instead of guessing.
• Profits fund clean energy and climate projects.

Let go of the guilt, keep the care

You don’t have to be perfect. Using AI thoughtfully, on a cleaner provider, for things that genuinely help you, is a reasonable choice. Push your workplace toward cleaner tools, ask providers where their power comes from, and spend your climate energy on the decisions that move the needle most.

Read more: the most eco-friendly AI chatbots and how much water AI uses.

Try Viro AI free.

Frequently asked questions

Should I feel guilty about using AI?

A single text answer uses very little energy, about as much as a TV running for a few seconds. What matters more is how much you use, what kind of tasks you ask for, and what powers the data center. Using AI thoughtfully, on a clean-energy provider, keeps the impact small.

How much energy does one AI question use?

Google measured 0.24 watt-hours for a median Gemini text prompt, and OpenAI says an average ChatGPT query uses about 0.34 watt-hours. Images, video, long documents and reasoning modes use much more.

What is the most eco-friendly way to use AI?

Use the smallest model that does the job, avoid unnecessary images and video, ask clearly once instead of regenerating, and choose a provider that runs on renewable energy. On recent Apple devices, an on-device model uses no data center at all.

Is AI worse for the environment than streaming or flying?

Per action, a text answer is far smaller than an hour of video streaming or any flight. The concern is scale: billions of AI requests a day, and fast growth in data-center electricity demand.

Sources

1. How much energy does Google’s AI use? We did the math, Google Cloud, August 2025
2. The Gentle Singularity, Sam Altman, OpenAI, June 2025
3. Key Questions on Energy and AI, International Energy Agency, 2026

Cover photo by Leyla Acar on Unsplash