Stop Blaming Yourself for AI’s Environmental Impact

On November 30th, 2022, ChatGPT was released to the public. Since then, the terms ‘artificial intelligence’ and ‘AI’ have been inescapable in everyday life. 

Regardless of your views, AI—and its seemingly endless integrations into every tech product released nowadays—is here to stay.

When most people use the term ‘AI’ or ‘Artificial Intelligence’ they are referring to large language models or LLMs for short. An LLM, like ChatGPT, is a neural network that is trained on vast swaths of data usually sourced from places like Wikipedia or books. Due to the “move fast and break things” nature of Silicon Valley and all the tech giants that seem to progressively control more and more of our lives, virtually everything an LLM could be trained on has already been incorporated into the training data for these AIs. 

Because of the relative recency of the AI boom, many concerns about the environmental impacts of AI have arisen.

To start off, AI data centers are terrible for the environment and consume far too much water. The people living nearby face higher utility prices, severe noise pollution, reduced water availability, and air pollution. An AI data center is a parasite that—once built—drains resources from and reduces quality of life for nearby residents. Unfortunately, the majority of AI data centers are built near rural communities—the communities most vulnerable to these problems. 

This being said, the blame is often unfairly shifted onto everyday users of AI which is a gross misunderstanding of reality. The real blame should fall on the giant corporations whose profits are prioritized over people.

To be concise, I will mainly focus on water consumption as an example.

According to a peer reviewed study, AI uses between 312.5 and 764.6 billion liters of water each year. According to the author, “That is of the same order of magnitude as all bottled water consumed worldwide in a single year.” (“AI’s Hidden Carbon and Water Footprint – Vrije Universiteit Amsterdam”)

That is a massive amount of water but it is important to know where exactly this usage is coming from and the actual specifics of water usage.

The vast majority of AI usage originates from private companies. Companies often have intensive AI agents that execute commands routinely that then cause more prompting on top of that. This adds up. Although there is no data on what exact percentage of data center usage comes from private companies, it is estimated to be more than 50%. Additionally, training new AI models—an extremely intensive process—takes up around another 10% of data center usage.

According to Google’s own data, one gemini prompt consumes 5 drops of water—or 0.26ml (Crouse). 

Let’s say a student performs 15 prompts a day for things like homework or help with essays—the actual number may vary, especially considering the inescapable google search AI.

Sam Altman—the ironically named CEO of OpenAI—has said that ChatGPT uses 0.32 ml of water per prompt, still just a few drops.  This estimate is accurate but only considers how much water the data center itself uses—not including the water used to generate the power. Including the water use generated from power plants, the 0.32ml is raised to an estimate of around 10ml of water (“AI Water Usage Per Prompt: What the Data Actually Shows”).

(Early research conducted in 2023 estimated that ChatGPT actually used 10-25ml per prompt but were conducted when AI models were much less efficient than they are now.)

Operating under this estimate, we can roughly estimate that the yearly water usage from AI would be 1.75L only taking into account water usage from the data center itself, and 54.8L including the water usage from power plants.

A single beef patty takes around 660 gallons—or 2,500 liters—of water to make (“How Many Gallons of Water Does It Take to Make a Hamburger?”). 

This means that a year of AI usage only accounts for ~2.2% of the water usage of a burger patty!

The point of this article is not to put others down for believing strongly in something that largely isn’t true; it’s to correct a viewpoint I see as misleading.

While, the total amount of water used by data centers is massive, a significant portion of it comes from corporate use/model training and—compared to industries like agriculture, which are in the tens, if not hundreds, of trillions of liters annually—AI water usage is actually relatively minimal.

Oftentimes, I see people react negatively to any mention of using AI as a tool—even shaming people who do use AI. It is as if Artifical Intelligence is some sort of Lovecraftian scourge brought upon humanity soley to destroy the environment.

In reality, AI use—even if you were an avid user—makes up a negligible percentage of your environmental impact.

All of this is to say that everyday people—like you and me—are not causing the very real negative impacts that come from AI. 

Instead, it is the massive corporations that prioritize profits over humans. 

Currently, AI models are being trained on how to best detect targets for drone strikes, analyze security footage to identify people, and find the best targets for missile strikes. 

The parts of AI that will actually cause meaningful harm to society do not come from people trying to find the best recipe or trying to figure how to solve a math problem but rather from the people whose goals align with infinite growth at the expense of the common good. 

In conclusion, you shouldn’t dislike the use of AI by everyday people but rather the corporations that use AI with complete disregard to the environment and human life.

Sources:

“AI Water Usage Per Prompt: What the Data Actually Shows.” Waterfreechat.Com, WaterFreeChat, 8 May 2026, https://waterfreechat.com/blog/ai-water-usage-per-prompt?utm_source=chatgpt.com. Accessed 14 May 2026.

“AI’s Hidden Carbon and Water Footprint – Vrije Universiteit Amsterdam.” Vrije Universiteit Amsterdam, 2019, https://vu.nl/en/news/2025/ai-s-hidden-carbon-and-water-footprint. Accessed 14 May 2026.

Crouse, Megan. “Google Claims One Gemini AI Prompt Uses Five Drops of Water.” TechRepublic, 21 Aug. 2025, \https://www.techrepublic.com/article/news-google-gemini-ai-environmental-impact-methodology/?utm_source=chatgpt.com. Accessed 14 May 2026.

“How Many Gallons of Water Does It Take to Make a Hamburger?” Biology Insights, 8 Jan. 2026, https://biologyinsights.com/how-many-gallons-of-water-does-it-take-to-make-a-hamburger/?utm_source=chatgpt.com. Accessed 14 May 2026.