---
title: "How can you reduce the environmental impact of using AI?\n10 actions ranked by effectiveness"
description: "Not all actions are equal: generating 2 times fewer videos avoids up to 26 times more than no longer saying thank you. 10 ways to reduce AI's impact, ranked."
canonical: https://projetcelsius.com/en/blog/reduce-ai-environmental-impact/
author: "Guillaume Pakula"
datePublished: 2026-10-05
dateModified: 2026-10-05
language: en
translationOf: https://projetcelsius.com/blog/reduire-impact-ia/
---

# How can you reduce the environmental impact of using AI? 10 actions ranked by effectiveness

**Author**: Guillaume Pakula
**Published**: 2026-10-05
**Section**: Explained
**Canonical URL**: https://projetcelsius.com/en/blog/reduce-ai-environmental-impact/
**French original**: https://projetcelsius.com/blog/reduire-impact-ia/
**Source**: Projet Celsius (https://projetcelsius.com/en/), a climate strategy consultancy of engineers and PhDs based in Paris and Marseille.

> Cutting back on video, choosing a smaller model or no longer saying thank you: for a single usage profile, we calculated what each of 10 actions avoids over a year. **3 of them can avoid more than 1 kg of CO2 equivalent (kgCO2e) per year**, while 2 avoid less than 50 g. The profile as a whole emits about **3 kgCO2e per year** from server electricity alone.

## Key takeaways

- Generating 2 times fewer videos avoids **300 g to 1.1 kgCO2e per year**, depending on the generator.
- Reasoning mode uses **1.9 to 30 times** the energy of a simple question, depending on the study and what it compares.
- An answer 2 times shorter uses **54% less energy**, on the model measured.
- Dropping 1 thank-you per day avoids **41 gCO2e per year**, as much as a single video emits.

Not all actions to reduce the environmental impact of using artificial intelligence (AI) are equally effective. For someone who generates 1 video per week, generating 2 times fewer avoids **about 1.1 kgCO2e per year** (1.1 kg of CO2 equivalent) with our reference generator, equivalent to **7.5 km by car**, or 300 g with a median generator. No longer thanking your assistant (1 thank-you per day) avoids **41 g**, which is 26 times less.

We quantified 10 actions for a usage profile that we chose: **10 questions per day, 5 images and 1 video of 5 seconds per week**. Of these, **3 actions matter, 5 help and 2 change almost nothing**. By our calculation, this profile emits about **3 kgCO2e per year** from server electricity alone, equivalent to **21 km by car**.

## The 10 actions to reduce the environmental impact of AI, from most to least effective

### How the actions are grouped

An action "matters" if it can avoid **at least 1 kgCO2e per year** and "helps" if it can avoid **at least 50 g**. Among the actions that help, running a model (the software that produces the answers) on your own computer only lightens the footprint if it is smaller than the model behind the online service.

### How it is calculated

For each action we assume a behaviour of our own choosing, which serves as its reference value: generating 2 times fewer videos, shortening 1 answer in 2, dropping 1 thank-you per day. **7 actions are based on a published measurement**; for the other 3 (whether AI is needed, follow-ups and stored conversations), none exists. Every figure in grams or kilograms in this article comes from our own calculation, using global average electricity.

**This profile is more active than average**: in July 2025, the average weekly active ChatGPT user sent **3.6 messages per day**, according to our calculation from the figures in a [study co-authored by OpenAI](https://www.nber.org/papers/w34255). No figure has been published for the number of videos generated per user.

**The ranking therefore depends on the profile.** For an intensive user, the total rises to **15 kgCO2e per year**, of which 11 kg comes from video, which stays at the top; **8 actions out of 10** can then avoid more than 1 kg. For an occasional user (2 questions per day, 1 image per week, no video), **reasoning mode moves to the top**: the total falls to **170 gCO2e per year**, and no reference value exceeds 35 g. Within a group, the order carries no meaning where the ranges overlap, and **the savings cannot be added together**.

## The actions that matter most: video, reasoning mode and whether AI is needed

### Generate fewer videos, and shorter ones: about 1.1 kgCO2e per year

In our profile, the 52 videos of the year emit **2.1 kgCO2e out of 3**, more than 70% of the total. This figure rests on a single generator: the most downloaded one on Hugging Face (a platform for open models, meaning models that can be downloaded freely), and **[the second most energy-hungry of the 7 (in French)](https://projetcelsius.com/ia/graphiques/video-generee-7-modeles/)** measured in the [study](https://arxiv.org/abs/2509.19222) it comes from. With the median generator, the one in the middle of the ranking (it produces lower-definition videos), generating 2 times fewer would avoid only **about 300 g**.

Duration matters even more than the number of videos: according to the same study, a video 2 times shorter needs **about 4 times less computation**, on the generator tested. [ADEME](https://librairie.ademe.fr/economie-circulaire-et-dechets/9495-avis-de-l-ademe-l-intelligence-artificielle-generative-des-impacts-environnementaux-importants.html), the French Agency for Ecological Transition, lists among its main principles of use for users that video generation must be limited and meet a specific need. Our article on [the energy used by an AI-generated video](https://projetcelsius.com/en/blog/ai-video-environmental-impact/) goes into detail.

### Do not leave reasoning mode on by default: from 0.46 to 15 kgCO2e per year

Reasoning mode makes the model work through [intermediate steps before answering](https://projetcelsius.com/en/ai/analyses/why-reasoning-and-agents-use-more-energy/): **it writes much more text**, and therefore uses more energy. PEReN, the French state's digital regulation expertise centre, measured open models for [Arcep](https://www.arcep.fr/uploads/tx_gspublication/rapport-IA-et-environnement_210526.pdf), the French electronic communications regulator: [with reasoning mode on (in French)](https://projetcelsius.com/ia/graphiques/mode-raisonnement-energie/), they use **92% more energy on average**, with an uncertainty range from -21% to +348%, and **849% more on average** on programming exercises.

The Hugging Face [AI Energy Score](https://huggingface.co/blog/sasha/ai-energy-score-v2) ranking compares 2 groups of models: those that reason consume **30 times more on average** than the others. For 3 models tested with and then without reasoning, it even measures **150 to 700 times more**.

This action is a special case: our profile includes no question in reasoning mode, so the action is **an avoided addition**. Switched on for the 3,650 questions of the year, this mode would add between 460 gCO2e (with the PEReN figure) and 15 kgCO2e (if we apply the factor of 30, although that compares 2 different groups of models). This is **the most uncertain item in the ranking**. The action itself is to **reserve this mode for the questions that call for it**.

### Ask yourself whether AI is necessary: from 0 to 3 kgCO2e per year

This is **the first condition of frugal AI** in the framework published by AFNOR, France's national standards body, as [summarised by Ecolab](https://greentechinnovation.fr/wp-content/uploads/2024/06/Referentiel-general-pour-lIA-frugale.pdf), a French state service: having demonstrated that AI is needed rather than a solution that consumes less. The gain is **that of the use avoided**, to which **we give no reference value**, only a range from nothing to the whole profile. Our comparison of [ChatGPT and a Google search](https://projetcelsius.com/en/blog/chatgpt-vs-google-search-energy/) quantifies the difference that a conventional search makes; with the reference values of this series, giving up a video avoids **as much as 300 questions would emit**.

## The actions that help: images, model size, short answers, complete requests and local models

### Generate fewer images, and smaller ones: about 170 gCO2e per year

The 260 images of the year emit **350 gCO2e** if we take the average of the generators measured. Generating 2 times fewer avoids **170 g**, and 2 times less again with a median generator. Size matters too: according to a [study from the University of Florence](https://arxiv.org/abs/2506.17016), an image 2 times smaller on each side needs **1.3 to 4.7 times less energy**, depending on the generator. See our article on [the footprint of a generated image](https://projetcelsius.com/en/blog/ai-image-carbon-footprint/).

### Choose the smallest model that does the job: about 140 gCO2e per year

According to a [study from Université Côte d'Azur](https://arxiv.org/abs/2510.01889), choosing the smallest sufficient model everywhere would reduce the energy consumed by AI in the use phase by **27.8%** overall. We apply this global average to our questions, for want of a better one. The gap between 2 models can be much larger: **small models specialised in one task** (such as summarising, translating or answering questions) consume **15 to 50 times less** than a general-purpose model according to [UNESCO](https://www.ucl.ac.uk/news/2025/jul/practical-changes-could-reduce-ai-energy-demand-90), although they do not suit every use. Our [ranking of ChatGPT, Claude and Gemini](https://projetcelsius.com/en/blog/chatgpt-vs-claude-vs-gemini-environmental-impact/) compares the models of the main providers. At Anthropic, [Fable 5.2 or Opus 5.5](https://projetcelsius.com/en/blog/fable-5-2-vs-opus-5-5/) shows what moving up to the most powerful model costs, and [how much energy and water Fable 5.2 and Opus 5.5 use](https://projetcelsius.com/en/blog/fable-5-2-opus-5-5-energy-water/) shows the effect of the effort setting.

### Ask for short answers: about 140 gCO2e per year

UNESCO and University College London measured the effect of length on an open model: [an answer 2 times shorter (in French)](https://projetcelsius.com/ia/graphiques/gestes-energie-evitee/) uses **54% less energy**, whereas a question 2 times shorter saves only **5%**.

On the same model run on another machine, writing a token (a word or part of a word) takes about **35 times more energy** than reading one, according to our calculation from the [measurements of Hugging Face researchers](https://arxiv.org/abs/2601.22357). We assume that 1 answer in 2 is shortened by half, which avoids **140 gCO2e per year**. ADEME also considers it **beneficial to specify the length or quality of the content you want**.

### Write one complete request rather than follow-ups: about 100 gCO2e per year

**Each follow-up makes the model write a new answer.** According to ADEME, precise queries **will require fewer repetitions and new queries**. We count **1 follow-up avoided for every 5 questions**, or **100 gCO2e per year**. This is an assumption, **as no measurement has been published**.

Attaching a long document costs more: according to an estimate by the research institute [Epoch AI](https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use), having the model read 100,000 tokens, or about 80,000 words, uses **about 40 Wh** (watt-hours), as much as 130 questions by our calculation. According to the same source, this cost is paid only once per conversation, so **attach only the pages you need**.

### Run a small model on your own computer

**With the same model, no gain has been established**: handling one request at a time, a computer chip and a chip in a data centre, the building where the servers run, use about the same energy, and a data centre card that handles 64 requests at once uses [12 to 20 times less per request](https://arxiv.org/abs/2511.07885). The gain comes from a smaller model: according to our calculation based on the [measurements in another study](https://arxiv.org/abs/2407.16893), a lightweight model on a laptop consumes **about 4 times less** than [a ChatGPT question](https://projetcelsius.com/en/blog/chatgpt-query-energy-consumption/), our reference, for an answer of equal length.

This gives a gain **from 0 to 380 gCO2e** per year, for the same electricity, according to our calculation. In France, electricity emits **11 times less** than the world average, 41 gCO2e per kWh against 458 g in 2025, according to data from [Ember](https://ourworldindata.org/grapher/carbon-intensity-electricity), an energy think tank: the balance becomes favourable if the online service runs on electricity close to that average. Our article on [AI presented as eco-friendly](https://projetcelsius.com/en/blog/most-eco-friendly-ai/) examines this case further.

## The actions that change almost nothing: thank-yous and deleted conversations

### Stop saying thank you: 41 gCO2e per year

In April 2025, a user wondered how much electricity people's thank-yous cost OpenAI. Its chief executive, Sam Altman, replied, as [the press](https://techcrunch.com/2025/04/20/your-politeness-could-be-costly-for-openai/) reported and without any calculation to back it up: "tens of millions of dollars well spent - you never know".

In January 2026, researchers at [Hugging Face](https://arxiv.org/abs/2601.22357) published a measurement on an open model: a thank-you consumes **0.245 Wh**, including the model's answer. 1 thank-you per day adds up to **41 gCO2e per year**, or 290 metres by car. **A year of daily thank-yous** uses about as much energy as a single video of 5 seconds, which equals about 370 thank-yous.

### Delete your conversations: less than 10 gCO2e per year

A year of questions and answers takes up around ten megabytes of text, and the generated images and videos a few hundred. Based on [the consumption of a data centre disk](https://www.cloudcarbonfootprint.org/docs/methodology/) and 3 copies of each file, keeping this data for a year takes **about 17 Wh** for the disk alone, which is **less than 10 gCO2e** according to our calculation and our assumptions. There can be other reasons to delete your conversations, such as protecting your privacy.

> **Outside the ranking: keep your phone for longer** A new smartphone has a footprint of about [80 kgCO2e](https://impactco2.fr/) over its life cycle according to ADEME, mostly from its manufacture: as much as **27 years of the server electricity of our AI usage profile**. The 2 figures do not cover the same scope, but the gap shows the order of magnitude. Our article on [the repair obligations placed on manufacturers](https://projetcelsius.com/en/blog/right-to-repair-manufacturer-obligations/) explains what they must make possible.

## In a company: choosing tools and setting rules of use

For an organisation, the [AFNOR](https://afnor.org/en/news/referential-for-measuring-and-reducing-environmental-impact-of-ia) framework on frugal AI, published in June 2024, lists **31 best practices**, grouped into 7 themes, which the Ecolab summary ranks by ratio of gain to effort, from best to weakest. One theme concerns the need: **checking that AI is a relevant way to meet it**.

Our guide explains how to [count AI in your company's Bilan Carbone®](https://projetcelsius.com/en/blog/ai-emissions-carbon-accounting-scope-3/) (the French carbon accounting method), another quantifies [AI's contribution to an SME's footprint](https://projetcelsius.com/en/blog/chatgpt-carbon-footprint/), and [our calculator](https://projetcelsius.com/en/tools/ai-carbon-footprint-calculator/) gives a first order of magnitude; to measure and manage usage over time, our article on [the providers that measure AI in companies](https://projetcelsius.com/en/blog/measure-ai-footprint-in-your-company/) compares the options. AI comes on top of computers, screens and phones, which our guide on [a company's digital carbon footprint](https://projetcelsius.com/en/blog/corporate-digital-carbon-footprint/) covers.

## What individual actions do not fix: missing provider data, default AI and data centre growth

### Measurements that providers do not publish

This ranking combines an estimate (the energy of a question), **researchers' measurements** made on open models, and our assumptions. For the most widely used assistants, there is **no measurement published by their providers** of the energy used by an image or a video, and PEReN did not test the reasoning mode of closed models such as ChatGPT's. Users therefore cannot check the effect of their own actions. The calculation also leaves out [the stages where AI has an environmental impact](https://projetcelsius.com/en/blog/why-ai-is-bad-for-the-environment/) before the answer: **chip manufacturing and model training**.

### AI switched on by default

Some AI is not something the user asked for: **answers generated at the top of search engine results**, and assistants built into software and phones. In its [July 2026 opinion](https://www.actu-environnement.com/media/pdf/news-48368-avis-ia-ademe.pdf), ADEME asks that, in traditional services, users be able to **switch off these features if they wish**. It also considers **training and awareness crucial** to using AI thoughtfully. To raise a team's awareness, our article on [choosing a speaker](https://projetcelsius.com/en/blog/ai-environmental-impact-speaker/) compares talk formats.

### The growth of data centres

One person's actions amount to **a few kgCO2e per year**. Data centres' electricity consumption rose by **17% in 2025**, an increase of **about 70 TWh** (terawatt-hours), according to the International Energy Agency (IEA): our article on [data centre electricity consumption](https://projetcelsius.com/en/blog/ai-electricity-consumption/) gives the orders of magnitude, and the one on [the 6 environmental impacts of AI](https://projetcelsius.com/en/blog/environmental-impact-of-ai/) looks beyond carbon alone.

## Key takeaways

- **3 actions matter**: generating fewer videos, and shorter ones; not leaving reasoning mode on by default; asking whether AI is necessary.
- **5 actions help**: fewer images, a smaller model, short answers and complete requests avoid from 100 to 170 gCO2e per year for those with a reference value, while a local model helps if it is smaller.
- **2 actions change almost nothing**: stopping the thank-yous (41 gCO2e per year) and deleting your conversations (less than 10 g).
- **This ranking applies to people who generate videos**: the profile emits about 3 kgCO2e per year from server electricity alone, more than 70% of it from video.

This ranking does not say whether you should use AI: our article on [the arguments for and against boycotting AI](https://projetcelsius.com/en/blog/should-you-boycott-ai/) addresses that question, and the one on [digital pollution beyond AI](https://projetcelsius.com/en/blog/digital-pollution-statistics/) places AI among other uses.

## Frequently asked questions

### Does saying thank you to ChatGPT waste energy?

Very little. According to a measurement by Hugging Face researchers published in January 2026, a thank-you consumes 0.245 Wh on an open model, including the model's answer; there is no published measurement for ChatGPT. 1 thank-you per day for a year adds up to 41 gCO2e with global average electricity, according to our calculation, or 290 metres by car. About 370 thank-yous use as much energy as a single generated video of 5 seconds.

### How can I use ChatGPT in a more eco-friendly way?

As with any assistant, 3 actions matter more than the others: if you generate videos, generate fewer and shorter ones; reserve reasoning mode for hard questions; and ask yourself whether AI is necessary. Next come fewer images, a smaller model and short answers. For our regular-use profile, the total is about 3 kgCO2e per year for server electricity alone, according to our calculation.

### What is frugal AI?

According to the framework published by AFNOR (France's national standards body) in June 2024, as summarised by Ecolab, an AI service is frugal if it meets 3 conditions: it has been demonstrated that AI is needed rather than a solution that consumes less; best practices are adopted by the producer, the supplier and the customer of the service; and the uses aim to stay within planetary boundaries. The framework lists 31 best practices.

### Does deleting your ChatGPT conversations save energy?

Not for the climate. Keeping a year of conversations, including generated images and videos, uses about 17 Wh per year for the disk alone, according to our calculation, which is less than 10 gCO2e. Deleting them can be justified to protect your privacy.

### What is the most effective way to reduce AI's environmental impact?

For people who generate videos, it is generating fewer videos, and shorter ones. With the generator used as the reference for this series, a video of 5 seconds emits about 41 gCO2e at the server, excluding the data centre, as much as 300 written questions, according to our calculation. In our profile, generating 2 times fewer avoids about 1.1 kgCO2e per year. Leaving reasoning mode on for all questions can have a larger effect: it uses from 1.9 times the energy of a simple question (average effect measured by PEReN) to 30 times (the average gap between models that reason and the others).

### Is it worth giving up AI to reduce your footprint?

On climate grounds alone, giving up all the usage in our profile avoids about 3 kgCO2e per year for server electricity, at the global average: the equivalent of 21 km by car. The question goes beyond climate: our article on boycotting AI reviews the other reasons put forward.

## Sources

- **Video, image and answer length**: [Delavande, Pierrard and Luccioni, Video Killed the Energy Budget](https://arxiv.org/abs/2509.19222), 2025, section 6; [Luccioni, Jernite and Strubell, Power Hungry Processing](https://arxiv.org/abs/2311.16863v3), 2024; [Bertazzini et al., The Hidden Cost of an Image](https://arxiv.org/abs/2506.17016), 2025; [UNESCO and UCL, Smarter, Smaller, Stronger](https://unesco.org.uk/site/assets/files/22732/394521eng.pdf), 2025, pp. 16 to 20 and Annex A.3, and [UCL press release](https://www.ucl.ac.uk/news/2025/jul/practical-changes-could-reduce-ai-energy-demand-90), 8 July 2025; [Delavande, Pierrard and Luccioni, Small Talk, Big Impact](https://arxiv.org/abs/2601.22357), January 2026; [Epoch AI, How much energy does ChatGPT use?](https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use), 7 February 2025.
- **Reasoning and model size**: [Arcep and PEReN, Generative artificial intelligence: what environmental challenges?](https://www.arcep.fr/uploads/tx_gspublication/rapport-IA-et-environnement_210526.pdf) (in French), May 2026, pp. 51 and 52; [Luccioni and Gamazaychikov, AI Energy Score v2](https://huggingface.co/blog/sasha/ai-energy-score-v2), Hugging Face, 4 December 2025; [Barros, Giroire, Aparicio-Pardo and Moulierac, Small is Sufficient](https://arxiv.org/abs/2510.01889), 2025; [Saad-Falcon et al., Intelligence per Watt](https://arxiv.org/abs/2511.07885), version of 6 September 2026; [Husom, Goknil, Shar and Sen, The Price of Prompting](https://arxiv.org/abs/2407.16893), version of 3 March 2026.
- **Uses, frameworks and recommendations**: [Chatterji et al., How People Use ChatGPT](https://www.nber.org/papers/w34255), NBER, September 2025; [ADEME, opinion on generative artificial intelligence](https://librairie.ademe.fr/economie-circulaire-et-dechets/9495-avis-de-l-ademe-l-intelligence-artificielle-generative-des-impacts-environnementaux-importants.html) (in French), 22 July 2026, p. 11, and [press release](https://www.ademe.fr/presse/communique-national/ia-generative-comment-quantifier-les-impacts/) (in French); [AFNOR Spec 2314, general framework for frugal AI](https://afnor.org/en/news/referential-for-measuring-and-reducing-environmental-impact-of-ia), June 2024, and the [Ecolab summary](https://greentechinnovation.fr/wp-content/uploads/2024/06/Referentiel-general-pour-lIA-frugale.pdf) (in French); [TechCrunch, Your politeness could be costly for OpenAI](https://techcrunch.com/2025/04/20/your-politeness-could-be-costly-for-openai/), 20 April 2025.
- **Equivalences**: [Ember, Global Electricity Review 2026](https://ember-energy.org/latest-insights/global-electricity-review-2026/), and [Our World in Data, based on Ember](https://ourworldindata.org/grapher/carbon-intensity-electricity), carbon intensity of electricity in 2025, life cycle basis (458 gCO2e per kWh worldwide, 41 in France); [ADEME, Impact CO2](https://impactco2.fr/), accessed 2 October 2026; [Cloud Carbon Footprint, Methodology](https://www.cloudcarbonfootprint.org/docs/methodology/), accessed 3 October 2026; [IEA, Key Questions on Energy and AI](https://iea.blob.core.windows.net/assets/3179f7f8-01f6-4dd6-bffa-c9f7b73f1dc9/KeyQuestionsonEnergyandAI.pdf), 16 April 2026, pp. 9 and 17.
- **Images**: Opening photo: Dayne Topkin, [Wikimedia Commons](https://commons.wikimedia.org/wiki/File:Lake_of_the_Woods,_United_States_%28Unsplash%29.jpg), CC0, cropped. Editing suite: Klaus Eichler / Max Mittelbach, [Wikimedia Commons](https://commons.wikimedia.org/wiki/File:Max_Mittelbach_digital_editing_suite.jpg), CC BY-SA 3.0 DE. Laptop: Solijon Solayev, [Wikimedia Commons](https://commons.wikimedia.org/wiki/File:Arch_Linux_system_update_via_pacman_on_an_Acer_laptop.jpg), CC BY-SA 4.0. Workshop: MarianneLC, [Wikimedia Commons](https://commons.wikimedia.org/wiki/File:Atelier_des_Sans_pagEs_%C3%A0_La_contemporaine_17_juin_2023_-_2.jpg), CC BY-SA 4.0, cropped.
