AI’s Hidden Footprint: The Pollution Behind the Promise

By: Chloe Ko

Artificial intelligence (AI), holds the recent spotlight for all the excitement we have for the future with its infinite potential. However, despite all the attention towards the bright side of AI, it is still important to consider the negative parts of AI that could largely impact us. 

When most of us think of AI, we probably imagine robots, or recently, smart assistants such as Chat GPT or Google’s Gemini. They are widely known and used by everyone due to their easy accessibility– often used to write stories, draw pictures, or do tedious assessments. And yes, AI can do all those amazing things. However what people aren't widely aware of is the darker side– its infrastructure, production, and use are creating real environmental problems. 

1.Energy Consumption and Carbon Emissions: 

One of the major concerns is the amount of electricity required to build, train and run the AI systems. According to the recent article from Massachusetts Institute of Technology (MIT) News, generative AI models with billions of parameters (“large models”) demand an insanely staggering amount of energy. For example, training a model like GPT-3, from Open AI, reportedly consumed 1,287MWh of electricity, which is enough to power ~120 U.S homes for a year. They also produced about 552 tons of CO2, which may easily build up to a larger environmental impact collectively, with thousands of other models are trained, retrained, and widely used. 

2.Water Usage & Other Resource Impacts: 

Beyond energy, AI’s backbone (data centers and hardware) uses tons of water and rare materials– both of which pose environmental risks. According to the United Nations Environment Programme (UNEP), data centers use water to cool their equipment to prevent them from overheating, and globally AI-related infrastructure “may soon consume six times more water than Denmark, a country of 6 million people” (UNEP). Furthermore, manufacturing of hardware like GPUs depend highly on rare earths and critical minerals, often minded in ways that are environmentally destructive. Data centers and associated hardware also generate electronic waste (e-waste) containing environmentally hazardous substances like lead or mercury. Even if AI might seem “virtual”, its physical footprint is very real and heavy. 

AI actually, is often hailed as a solution for climate change, environmental monitoring, biodiversity protection, etc. Even though it has the potential ability of helping those things, it is more important that we don’t keep ignoring its environmental costs while chasing those benefits. We won’t want to end up undermining the very planet that technology is supposed to protect. As UNEP notes, “we need to make sure the net effect of AI on the planet is positive before we deploy the technology at scale”. 

So next time you use an AI tool to write a poem, draw a picture, or get help with homework, remember: behind the screen, there’s a large physical infrastructure with real environmental costs. The excitement around AI is understandable—but the environmental side deserves attention too. Us, as part of members on this earth, who thrives from its borrowed resources, we can ask ourselves: Is this technology being developed responsibly? Are we accounting for its hidden costs? And how can we push for innovation and sustainability at the same time?

 AI’s promise is real—and so is its pollution. By talking about both, we can help steer technology toward a future where it helps people and the planet, not just one.