ChatGPT, Gemini, Grammarly, Canva, Zapier: If you’ve been off the grid (and a two-finger salute to you, if you have), you might mistake these for new X-Men characters hitting the market, but in fact they are just some of the more popular AI tools dominating the virtual space today.  

And it’s not difficult to see why.  Engaging with a large language model (LLM) like ChatGPT can be instructive, fun, educational.  It’s a boredom-slayer while you sit around waiting to hear your name called at the doctor’s office.  It throws cute little emojis your way, bolsters your confidence with unearned compliments, and in the most antiseptic, clinical prose, will even lay out a speech before you.  

On anything.  

And instantly.

But what transpires within that ‘instant’ is not really well known.  According to MIT News, a single text query on ChatGPT consumes about five times more electricity than a simple web search.  Now multiply this by roughly 2.5 billion search prompts daily, and you see where this is headed.

The sobering reality is that AI is here to stay, and its immersive ubiquity will be such that future generations will recognize it no more than fish recognize the water they swim in.  Whether you find this prognosis sad or cause for celebration will, of course, depend on you, but what can’t be disputed is the impact of AI on the environment.  Let’s take a look.  

The Impact of AI on the Environment

The High Energy Demands of Data Centers

The generative impact of AI on data centers is astounding.  Researchers report that North American data centers doubled their power usage in one year, going from 2,688 megawatts in late 2022 to 5,341 by the end of 2023, an uptick largely attributed to generative AI.  Globally speaking, data centers — and, as we know living in Northern Virginia, Ashburn is at the center of this —  consumed 460 terawatt-hours in 2022, and projections for the future are, well, bleak.  By 2026 this number will have risen to 1,050 terawatt-hours annually.

 

Cooling Data Centers

While the generative impact of AI on data centers may get some coverage, their heavy reliance on water for cooling is another factor with real environmental consequences.  The heat generated by so much hardware and equipment in data centers is cooled by using water, and for every 20 to 50 questions you might ask ChatGPt about dinner recipes or hives or toenail fungus, it will consume roughly the equivalent of a 500ml bottle of water.

Worse yet, that demand for fresh water could outpace supply by 40% by the end of the decade.    

 

Carbon Emissions and Large Language Models

And what about carbon emissions?  Google’s sustainability report for 2024 reported that its greenhouse gas emissions rose 48% since 2019, attributing the inflation to its data center energy consumption as well as supply chain emissions.  Because large language models require training, and are like immense energy-swallowing engines fueled by electricity, the servers powering them release carbon into the atmosphere at disastrous rates.

As if this wasn’t bad enough, once trained, models like ChatGPT still burn through energy during inference, which is basically the process of producing answers to the questions you ask.  And because the shelf-life of AI models is relatively short, with companies rolling out newer models every few weeks, the energy used to train prior versions is essentially wasted.

 

AI’s Growing E-Waste Problem

The impact of AI on the environment isn’t confined to electricity and water alone, however.  AI relies on hardware, GPUs, servers, chips, that time out quickly, and tech companies are constantly upgrading.  Which ties into the ever-present problem of e-waste, its continual global rise, and what to do with these outdated electronics (more on this, here).

What it boils down to is this: without properly recycling and circular economy efforts, AI will worsen toxic waste streams and intensify pressure on mining operations, operations that often come at the cost of fragile ecosystems and, yup, exploited workers.

 

Not All Doom and Gloom

Now, it’s not all doom and gloom, however.  People can readily see the impact of Air on the environment through a negative lens, but AI also offers many positives.  Apart from its well known uses, according to Associate Professor of Industrial Ecology and Sustainable Systems, Yuan Yao, many AI applications can bring environmental benefits as well.  Among them: better energy efficiency, lower energy usage, as well as render the almighty assist in environmental monitoring.  All good things, no doubt.

But how?  Essentially by applying AI the way most of us do to handle what daily life throws our way, like a tool, only in this case to improve life cycle assessments, measuring a product’s environmental impact from beginning to end.

 

Simple Ways Everyday People Can Reduce AI’s Environmental Impact

Undoubtedly, the real fight to reduce the impact of AI on the environment is with the tech industry at large.  From prioritizing algorithmic efficiency to adopting carbon-aware scheduling to choosing renewable-powered data centers, the list is quite extensive.

That being said, what can simple, everyday people do about it?

A few things, actually.

 

1. Recycle old devices responsibly, always using a certified recycler like yours truly.

 

2. Get involved and advocate for transparency, urging AI companies to disclose their sustainability reports.

 

3. Turn off unused apps or tabs, a small but helpful gesture that keeps devices from using more power.

 

4. A follow up to this, log out of AI apps to prevent background use and energy usage.

 

5. Limit any frivolous questions or prompts.  A little self-restraint goes a long way.

 

6. Promote a circular economy by buying used electronics, and keep outdated devices in the market stream.

 

7. Be selective, and avoid generating images, videos, text or music simply to entertain yourself.

 

AI Is Here To Stay

The subtitle says it all, and there’s no way around it.  It’s a brave new world, and what the future holds remains to be seen by those who will live it.

We are living our time, however, have agency, and can exert more self-discipline to ensure those to come have access to the marvelous riches with which the world abounds.  It is a gift.

Staying informed, limiting AI usage, avoiding unnecessary search prompts, recycling e-waste — all these serve a vital function towards a greener, more sustainable tomorrow.