AI & the Environment: What You Need to Know –and How to Make a Difference

AI has become an increasingly important topic, given its recent boom. If you are like me, based in the Bay Area for most of the year, you know it feels like there is another AI startup every day. Over the last few years, generative AI has exploded across industries, schools, creative tools, and even our personal lives. AI platforms can now write essays, generate images, help debug code, draft emails, or even help you look for recipes and make a grocery list. OpenAI now reports up to 2.5 billion prompts sent to ChatGPT’s models daily. Everyone is scrambling to integrate AI into their daily life for efficiency’s sake, and some are even prompting its use unknowingly within a browser search.

However, many people are unaware that AI does not function solely on code hidden away on a single server. To work at the speed we have come to expect, it needs water, electricity, and infrastructure to sustain the constantly updating servers. Therefore, the future of AI does not depend on the next big startup; it is a question of sustainability. But who says that AI won’t be what helps us solve the climate and energy crisis? So, we must learn to use it responsibly.

Many people don’t know exactly what happens behind a search on ChatGPT or any other AI server. In simple terms, it triggers a massive computational process in a data center, scouring huge networks of information and using significant amounts of energy. Computer science advancements have made this faster, but not less energy-intensive.

The scale is staggering. A UC Berkeley Lawrence Laboratory study found that in 2024, data centers powering AI models used 200 terawatt-hours of electricity, almost enough to power Thailand, or about 7.2 million U.S. homes for a year. Researchers predict that the energy expenditure could rise to 326 terawatt-hours per year by 2028. That can power 22% of U.S. households each year. This doesn’t have to be bad. If data centers were sustained by renewable energy, the impact would be far smaller. But, in regions dependent on fossil fuels, the rapidly expanding data centers could produce the same emissions as driving over 300 billion miles. And that still doesn’t account for the transportation and energy needed to build these centers. 

Between 2024 and 2028, the share of US electricity going to data centers may triple, from its current 4.4% to 12% while major companies are projected to spend 500 billion dollars towards building new data centers for AI.

Then there’s the water. Freshwater is used to cool the computers running AI models, via circulation in the data centers. MIT News notes that generally, each kilowatt-hour of energy a data center consumes typically requires about two liters of water for cooling. Another statistic from a 2023 study at UCR found that Microsoft’s data centers consumed about 700,000 liters of freshwater, about the same amount of water used in the manufacture of about 370 BMW cars, just to train GPT-3. That usage is simply production cost, not accounting for daily interaction.

By now, you likely have a pretty clear picture of AI’s resource demands and its drawbacks. Yet it's also true that AI has become a powerful tool in most industries, especially healthcare. Many companies point to AI’s potential to help us solve climate challenges, even while expanding their own water and energy use. This review is meant to help you use AI and trust it, to be greener.

When it comes to climate issues, it can feel like one person’s choices pale in comparison to corporate action, yet here are a few ideas for your One Green Thing, so that you can use AI responsibly:

1. Use Low-Compute Tools When Possible

If you are looking for grammar checks and quick searches, avoid the large language models and turn off your automatic Google Gemini review. Simple searches have sustained us for decades, and think of it as a chance to exercise your problem-solving skills!

2. Prompt with Purpose

Many users tend to send multiple small prompts, revising and re-prompting repeatedly. Before you hit enter, put all your questions in one clear prompt, and ask the model to give you a summary of its process. Drafting your own outline or source material first lets the AI deliver useful information in one step, saving both your time and the planet’s resources.

3. Support Sustainable AI Providers

Some AI platforms invest in renewable energy for data centers and recycle cooling water. Even tech giants like Google and Microsoft have committed to carbon-negative goals by 2030. Research your tools and support companies whose actions align with your beliefs and their environmental claims. 

Finally, I’m concerned not only about AI’s environmental impact, but also its mental one. I urge you to examine how you use AI and consider your own dependency. AI hasn’t been around long enough for researchers to fully measure its cognitive effects, but the temptation toward intellectual laziness is clear. That being said, let’s use AI as the tool that it is, while doing our minds and the environment a favor!