A teacher friend recently showed me at Starbucks how he prepares quizzes these days. No stacks of books. No late-night writing sessions. Just a laptop, an AI tool, and a cup of coffee. By the time the brew was warm enough to enjoy, the first draft was finished.
Around the same café shop, similar scenes unfolded. One student was asking a chatbot to explain a difficult lesson in chemistry. Another was generating reviewer notes. A friend was drafting a speech. Beside me, my daughter was reviewing her pediatrics notes. Nobody seemed amazed anymore. AI has become like Wi-Fi. We only notice it when it is unavailable.
Recently, while waiting for a meeting, I watched people around me navigate life through screens, apps, and devices. One person consulted AI. Another paid digitally. A smartwatch quietly vibrated. No one appeared impressed. Convenience had become ordinary. Neither was I. Technology has become as routine as turning on a light switch. Yet while watching that scene unfold, I realized how rarely we think about the invisible machinery making it all possible. We marvel at the convenience. We seldom ask about the environmental bill arriving quietly in the background.
Ironically, public attention focused on a detail almost too small to notice. Some reports suggested that extra words in prompts require additional computing resources. Before long, people were joking about whether saying “please” and “thank you” to ChatGPT was hurting the planet.
Personally, I hope courtesy survives the digital age. The issue is not politeness. The issue is scale.
One polite prompt means almost nothing. One student generating a reviewer does not matter. One teacher asking AI to prepare a rubric will not trigger an environmental crisis. But multiply those actions by hundreds of millions of users every day and the picture changes. The same principle applies to electricity at home. Leaving one light bulb on is not a problem. Leaving millions of light bulbs on certainly is.
Part of the challenge is that digital life feels weightless. We upload files to “the cloud.” We stream videos. We join online meetings. We ask AI to summarize reports, generate images, or explain concepts. Everything appears instant and invisible. The language itself is misleading. Clouds sound soft. Harmless. Almost magical.
The reality is less poetic.
Behind every AI-generated answer is a hidden world most users never see. Data centers operate around the clock. Electricity flows. Water cools equipment. The cloud is not weightless. It has a footprint.
That reality reminded me of the lockdown years when our homes suddenly carried the load of schools, offices, and social gatherings. Many families felt the impact immediately through rising power bills. What seemed invisible eventually showed up on paper. Internet subscriptions and pisonets became necessities rather than luxuries. Parents who once worried about school supplies found themselves budgeting for gadgets and data plans instead.
Technology solved many problems. It also revealed that convenience is never completely free. Artificial intelligence follows the same pattern.
Researchers from MIT have noted that generative AI systems consume significantly more computing power than traditional searches because they create responses rather than merely retrieve information. The United Nations University recently emphasized that AI should not be viewed only as software but as a material system requiring electricity, water, land, and supply chains. Every answer has a footprint, even if users never see it.
The scale is easy to underestimate. According to projections by the International Energy Agency, data centers worldwide could consume close to 945 terawatt-hours of electricity annually by 2030, much of it driven by artificial intelligence. That is roughly equivalent to the current yearly electricity consumption of an industrialized nation like Japan. Put differently, if AI-related data centers were a country, they would rank among the world’s largest electricity users. Suddenly, a simple chatbot response feels connected to something much bigger than a screen and a keyboard.
The water issue may be the most surprising. Most people understand that computers need electricity. Fewer realize that they also need cooling. The internet may feel weightless, but it is not. Servers run hot, and cooling them requires water—lots of it. In some places, the water used to cool data centers could fill thousands of household water tanks each day. For families who have stored water in drums during shortages, that reality is easier to understand than most technical reports.
The challenge, then, is not choosing between technology and the environment. It is learning how to enjoy the benefits of one without quietly exhausting the other. We want the opportunities technology brings, but we must also recognize the resources it consumes.
That does not mean AI is the enemy. It is already helping communities prepare for disasters and better understand environmental threats. The question is not whether to use it, but how to use it responsibly.
Perhaps that is why this conversation matters. The challenge is not choosing between technology and sustainability. It is recognizing that one ultimately depends on the other.
Every convenience leaves a trace. Every connected device has a backstory. Every prompt depends on a chain of resources stretching far beyond the screen in front of us.
Perhaps the real measure of progress is not how intelligent our machines become. Perhaps it is whether we remember what makes that intelligence possible.
Because long before there were algorithms, there was water. There was energy. There was a living planet sustaining it all. And no amount of artificial intelligence will matter if we fail to protect the very intelligence of nature that makes all human innovation possible.
Doc H fondly describes himself as a “student of and for life” who, like many others, aspires to a life-giving and why-driven world grounded in social justice and the pursuit of happiness. His views do not necessarily reflect those of the institutions he is employed or connected with.
