The Untold Harm: The Environmental Impact of AI

Written by Emir Taha Macit

AI and its countless models have most certainly become a significant part of many people’s daily lives. Whether it is for making plans, researching, or just chatting, AI is a tool used very frequently. This frequency is what brings forth a major problem in the system of how AI works: environmental concerns. AI’s usage of several resources, such as electricity and water, and its carbon emissions make it a highly polluting and wasteful tool for the environment. These concerns led to the emergence of advancements in terms of sustainability. Despite how useful AI is, there is no understating its environmental harm; thus, it should be used in moderation.

The first thing we will be going over is AI’s use of electricity and the carbon emissions it generates. To power the vast amount of electronics across the large areas that data centers cover, to provide their services to users worldwide, and to train the AI in the first place, developers need a lot of energy. Sadly, most of this energy comes from fossil fuels, leading to carbon emissions comparable to those of some nations, given the ever-increasing number and activity of data centers. According to some experts, researching something via AI models uses 5 to 10 times the electricity needed to search it up on Google. This piece of information is nothing but a highlight of just how much energy is used and carbon is emitted through AI. The electronics that are used, however, have a problem of overheating when unattended. Which leads to the next problem.

The use of water to cool down electronics in AI’s data centers is another major harm of AI to the environment. The continuous use of this hardware leads to heating, requiring a large amount of water to cool. The reduction of water that was taken from natural sources negatively affects ecosystems and lessens the already-lowering supply of drinkable water that the world has to offer. Furthermore, the water used for AI could have been used for something more eco-friendly and humanitarian, such as agriculture. But water is not the only natural resource used for AI.

To develop and train an AI model, advanced hardware is a necessity. The production of this hardware requires various materials from the earth, such as lithium and cobalt, along with some elements classified as rare earth elements (REEs), such as neodymium and palladium. While they do not cause any problems once the process of hardware production is done, their acquisition raises environmental concerns. Mining for some of these materials has its own problems with water usage, along with pollution. Drills and heavy machinery release copious amounts of CO2 into the atmosphere, adding to the destructiveness of the whole process of making and maintaining AI.

Another harmful product of the processes of AI, other than greenhouse gases, is e-waste. E-waste is any part of an electrical device, or a device itself, that is no longer useful. This includes components like GPUs and old servers, components used in AI data centers. E-waste is especially problematic due to several reasons. These include, but are not limited to: the difficulty of their disposal or storage, the toxic materials they contain, and the awkwardness of their recycling due to the many mixed materials. These problems are made even worse by the quick advancements in hardware production, leading data centers to frequently replace the old ones with new ones.

All of the environmental concerns of AI that we have gone over have been recognized in some way. Advancements in the efficiency of the energy and water used in data centers have made them more sustainable, partially solving some problems and showing us that AI may actually become a relatively sustainable tool if worked on. Another advancement is that the development and training of AI models have become increasingly more efficient and quicker, requiring fewer resources overall. However, this is not to say AI is or will definitely become eco-friendly. Merely for the foreseeable future, data centers will keep using resources that harm the environment to some degree.

In conclusion, AI is an environmentally hazardous tool that needs to be used with consideration for the harm inflicted during its production and development. Despite the assistance it may provide, it raises too many environmental concerns to ignore. As new advancements are made, AI may become a reasonable tool to use without even considering any ecological problems.

References:

1. “The environmental impact of AI and how to mitigate it”, PwC Belgium, 2025, https://www.pwc.be/en/news-publications/2025/responsible-ai-environmental-impact.html#:~:text=The%20most%20scrutinised%20environmental%20impact,and%20data%20centre%20in%20question.

2. Adam Zewe, “Explained: Generative AI’s environmental impact”, MIT News, 2025, https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117

3. “AI has an environmental problem. Here’s what the world can do about that.”, United Nations Environment Programme, 2025, https://www.unep.org/news-and-stories/story/ai-has-environmental-problem-heres-what-world-can-do-about

4. Shaolei Ren, Adam Wierman, “The Uneven Distribution of AI’s Environmental Impacts”, Harvard Business Review, 2024, https://hbr.org/2024/07/the-uneven-distribution-of-ais-environmental-impacts

5. Robert A. James, Ashleigh Myers, “AI Needs Critical Materials, Fast! But From Where?”, Gravel2Gavel, 2025, https://www.gravel2gavel.com/ai-critical-materials/#:~:text=AI%20hardware%20relies%20on%20familiar,of%20data%20at%20high%20speeds.

6. Sammy Witchalls, “The Environmental Problems Caused by Mining”, Earth.org, 2022, https://earth.org/environmental-problems-caused-by-mining