While the increasing role of generative AI in our everyday lives has brought many conveniences and opportunities never thought possible, its detrimental impact on the environment has gone drastically underreported. As often happens with new technologies such as this, we’ve been all too happy to enjoy the newfound convenience and ease while ignoring the harm done to provide us this service. While the controversy surrounding AI’s intensive water use and the stress it puts on electricity infrastructure are far from obscure, the true extent of its environmental impact is scarcely fully understood. Too often, these issues are summarized in a single word or phrase, simply brushing on the subject before moving on to the more interesting and optimistic elements of the technology. With 92% of students in 2025 reporting employing the use of generative AI to complete assignments and with AI becoming more integrated into curriculum itself, these issues are more relevant than ever. Therefore, I feel it’s important to give a more comprehensive overview of the true damage AI can cause so that students can make more informed decisions on when and if to use it.
One of the most well-known and largest sources of environmental damage from generative AI is its water use. While most people access tools like ChatGPT or Google Gemini from their laptops or phones, the actual processing and operation of those AI models is handled in giant buildings called datacenters filled with powerful computers, which take your prompt and perform complex and power-intensive calculations to return an answer. Though these computers are extremely efficient, they still lose much of the power they use to heat. Because these computers operate best within a certain temperature range, they need somewhere to vent this heat. That’s where the water comes in. Datacenters draw massive amounts of fresh water from their local area to cool their servers, with the largest among them using an average amount roughly equivalent to 4200 people. Despite the fact AI is a relatively new industry, these datacenters rank in the top 10 most water-consuming industries. This scale of water consumption can exacerbate already existing water shortages in many water-stressed areas. All this considered, it’s not hard to see why many researchers are extremely concerned about AI’s water use.
Water use isn’t the only environmental concern from datacenters, either. These facilities also use massive amounts of electricity to power their servers. In many cases, local power infrastructure is simply not sufficient to sustain these operations. In these cases, AI companies often build natural gas turbine generators to supplement or replace local power. These gases emitted by these turbines not only contribute to global warming, but also have catastrophic effects on the residents of local communities. In areas in Memphis, Tenn., near an xAI datacenter employing the use of 35 gas turbines, cancer rates are four times the national average, with many residents also complaining of respiratory issues since the datacenter started operation.
As with most environmentally harmful industries, the damaging environmental effects of this and other datacenters falls disproportionately on minority communities in a phenomenon called environmental racism. AI companies often specifically target these communities for new operations because they lack the political power to resist them, giving the companies an opportunity to build and operate datacenters for cheap.
While one article can’t possibly cover every aspect of the discriminatory and environmentally destructive practices and processes undertaken by AI companies, I hope this brief dive into a few of the many ways AI harms people and the environment will help students make better decisions when it comes to using AI. I understand it can be extremely difficult to cope with the workload in college courses. AI can be an extremely tempting tool to lighten the load, and with courses being increasingly designed around the idea that students will use AI to complete assignments, the workload is ever-increasing. It would be unrealistic to ask that students cease AI use altogether, but it’s my hope that after reading this article, you’ll be more likely to consider the very real consequences of your AI use and to moderate how and when you turn to AI for help. AI is no doubt here to stay, but if we all stay conscious of its effects and use it only when necessary, progress can be made towards a world where AI can be the force for good we all want it to be.