AI chatbot carbon footprint: AI Chatbots’ hidden cost: How your questions harm the planet through carbon emissions

AI chatbot carbon footprint: Each time you ask an AI chatbot a question, it’s not just giving you an answer,it is also using energy and producing carbon dioxide (CO₂), that harms the environment.

AI Models That “Think” More Emit More CO₂

Researchers in Germany found that some AI models, especially those that “think” a problem by generating long, step-by-step reasoning before answering, can emit up to 50 times more CO₂ than models that give short, direct answers, as per a report. Surprisingly, this extra reasoning doesn’t always lead to better answers.

AP

Carbon Emission (AP Photo/Joshua A. Bickel)

Why Do AI Tools Generate Emissions?

How does this happen? Well, AI answers are made up of tokens, these are words or parts of words converted into numberical data, which the AI can then process, as reported by Scitechdaily. Generating these tokens, along with the computing power needed, results in CO₂ emissions, according to the report. As this happens in the backend and can’t be seen, most people don’t realize that using AI tools actually comes with a noticeable carbon footprint, as per the Scitechdaily report.

iStock
ALSO READ: MrBeast says he bought NFL – here’s the truth behind the viral video

Live Events

German Study Analyzes Emissions from 14 AI Language Models

In a study published in ‘Frontiers in Communication’, Maximilian Dauner and his team at Hochschule München University of Applied Sciences tested 14 large language models (LLMs) with sizes ranging from 7 to 72 billion parameters (parameters determine how a model learns and makes decisions) using 1,000 standard questions from different subjects, according to the report.

ALSO READ: Is AI making doctors lazy? Study reveals overreliance may be undermining their critical skills

AI chatbot energy consumption

High AI Accuracy Model Comes at a Higher Environmental Cost

They discovered that models designed for detailed reasoning produced on average 543.5 “thinking” tokens per question, which was way more than the 37.7 tokens from models that keep answers short. More tokens mean more CO₂, but that doesn’t always improve accuracy, as per the Scitechdaily report. These thinking tokens are the extra internal content generated by the AI model before it settles on a final answer and the extra detail may not improve the accuracy of the answer but it does increase the environmental cost, as reported by Scitechdaily.

The study found that the best accuracy was seen in a reasoning model called Cogito with 70 billion parameters, scoring 84.9% in accuracy, according to the report. But it also produced three times more CO₂ than similarly sized models that kept answers concise, as per Scitechdaily.

iStock
Dauner explained that there’s a clear trade-off between accuracy and environmental impact as none of the models that kept emissions under 500 grams of CO₂ equivalent (CO2 equivalent is the unit used to measure the climate impact of various greenhouse gases) reached more than 80% accuracy, according to the report.

He said, “Currently, we see a clear accuracy-sustainability trade-off inherent in LLM technologies,” adding, “None of the models that kept emissions below 500 grams of CO₂ equivalent achieved higher than 80% accuracy on answering the 1,000 questions correctly,” as quoted by Scitechdaily.

The subject matter also made a difference in the levels of CO2 emissions. Questions needing deep reasoning, like abstract algebra or philosophy, caused up to six times more emissions than simpler subjects like high school history, as reported by Scitechdaily.

ALSO READ: Dirty Coffee is brewing up a new global trend! Here’s how to make this 2-ingredient drink at home

How to Reduce AI’s Carbon Footprint with Smarter Prompts

With the findings of the study, the researchers have said that they hope their work will cause people to make more informed decisions about their own AI use, according to the report. They suggest prompting AI for concise answers and reserving heavy-duty models for tasks that truly need them, as per the Scitechdaily report.

Dauner recommended, “Users can significantly reduce emissions by prompting AI to generate concise answers or limiting the use of high-capacity models to tasks that genuinely require that power,” as quoted in the report.

Hidden cost of AI chatbots

The researchers have even pointed out that the choice of model can also make a significant difference in CO2 emissions, according to the Scitechdaily report. For example, answering 600,000 questions with the DeepSeek R1 model (70 billion parameters) emits as much CO₂ as a round-trip flight from London to New York, as per the report. While, the Qwen 2.5 model (72 billion parameters) can answer about 1.9 million questions with similar accuracy while producing the same emissions, as reported by Scitechdaily.

However, the researchers pointed out that their results may be impacted by the choice of hardware used in the study, an emission factor that may vary regionally depending on local energy grid mixes, and the examined models, so these factors may limit the generalizability of the results, according to the report.

Dauner said, “If users know the exact CO₂ cost of their AI-generated outputs, such as casually turning themselves into an action figure, they might be more selective and thoughtful about when and how they use these technologies,” as quoted by Scitechdaily.

ETtech
ALSO READ: Top AI Tools of 2025: Is ChatGPT still leading or is Gemini, Grok, DeepSeek taking over?

FAQs

Do AI chatbots really harm the environment?
Yes, they consume energy and produce CO₂, especially when they generate long, complex responses.

Can I reduce AI emissions just by changing how I ask questions?
Yes, prompting AI to give concise answers can significantly reduce emissions.

Add as a Reliable and Trusted News Source



_{Images are for reference only.Images and contents gathered automatic from google or 3rd party sources.All rights on the images and contents are with their legal original owners.}