Your Dumb AI Prompt Is Not Harmless. The 1980s Trend Has a Water Cost

Yes, yes, yes. You all look beautiful today. Apparently, all of you looked gorgeous in the 1980s too, and judging by the latest AI trend, all of you are going to look even hotter in 2050. There is just one small problem. You are asking some of the world’s most powerful computing systems to manufacture imaginary versions of yourselves for a few seconds of social media validation.

Type a prompt asking ChatGPT or another AI model to turn you and your partner into an old-school 1980s photograph, wait for the servers to do their work, post the result, collect a few hearts and move on. It feels harmless because the transaction is invisible. You see a text box, not a warehouse full of computing equipment. You see an image, not the electricity and cooling infrastructure behind it. Your individual request will not drain a reservoir, of course. That is not the argument.

The problem is what happens when millions of people make the same supposedly harmless request, repeatedly, simply because a trend appeared on their feeds. We have become remarkably good at demanding convenience while remaining completely detached from its cost. AI is not magic. Every generated image requires computation, and computation requires infrastructure. That infrastructure requires energy, water and materials. The fact that you cannot see any of it from your phone does not mean it is not there.

AI Was Built to Solve Problems. Not Manufacture Validation

There is an important distinction being lost in the current AI frenzy. Artificial intelligence can be enormously useful. It can automate repetitive work, assist researchers, help programmers, translate languages, improve accessibility and give artists powerful new tools. It can save people hours and, in some cases, unlock work that would previously have been impossible. Those are compelling reasons to use it.

Asking a machine to generate yet another picture of you looking younger, richer, cooler or like you belonged in a 1980s family album is a different proposition. The environmental footprint of AI is complicated, and anyone claiming that every prompt consumes exactly a particular amount of water is oversimplifying the science. Consumption varies according to the model, hardware, data centre, cooling system, location and workload. But the broader issue is real.

A widely cited study estimated that training GPT-3 in Microsoft’s US data centres could directly evaporate around 700,000 litres of freshwater. Researchers also projected that global AI demand could be associated with 4.2 to 6.6 billion cubic metres of water withdrawal annually by 2027. Those numbers are estimates, not a meter attached to your individual prompt. But they illustrate the scale of a technology that is rapidly becoming embedded in everyday life.

The uncomfortable question, therefore, is not whether one silly image will destroy the environment. It is whether we should normalise the idea that every fleeting curiosity deserves computational resources simply because the technology makes it easy.

We Lecture People About Sustainability, Then Waste AI for Fun 

The contradiction becomes especially difficult to ignore when the people participating in every AI fad are otherwise deeply concerned about sustainability. We are told to switch off lights, reduce plastic, waste less food, conserve water and make more responsible choices. Then an AI trend arrives and suddenly generating twenty versions of the same photograph becomes entertainment.

Generate it. Change the clothes. Generate again. Make it younger. Generate again. Add a partner. Change the background. Generate again. Delete it. Start over. The cycle is almost comically wasteful. And the worst part is that much of it is not even creativity. It is insecurity dressed up as technology. We want to know how we would have looked at 25. We want a cinematic version of our wedding. We want an AI portrait in Paris, another as a celebrity, another as a billionaire and another as a character from a movie.

There is nothing inherently wrong with having fun, but we have reached a point where the ability to generate something has been confused with a reason to generate it. You do not have to participate in every trend that appears on your screen. You do not have to ask a machine to manufacture a more attractive version of yourself. And you certainly do not need an algorithm to validate memories you already possess. If you actually want to know what your parents looked like in the 1980s, open an old photograph. It may be faded, badly exposed and completely imperfect. That is precisely why it matters.

The Real Problem Is Not One Fake 1980 Picture. It Is What Comes Next

The larger issue is not one photograph. It is the culture we are building around AI. Technology companies are racing to make AI cheaper, faster and more powerful, while consumers are being encouraged to use it for increasingly trivial tasks. At the same time, governments and communities are confronting the energy and water demands of expanding data-centre infrastructure.

In the United States, the Lawrence Berkeley National Laboratory estimated that data centres consumed about 66 billion litres of water directly in 2023 and projected direct consumption could rise to roughly 150 to 280 billion litres annually by 2028, depending on how the industry develops. The figures vary, the technology changes and efficiency will improve. But the direction is unmistakable: AI is becoming a major consumer of physical resources even though the product appears entirely digital. That makes our choices worth examining. This is not an argument to abandon AI. It is an argument to stop behaving as though computation is free.

Use AI when it solves a real problem. Use it when the benefit is meaningful. Use it when it helps you create, discover, work or learn. But when the only objective is to generate another disposable picture for five seconds of social media attention, perhaps the smartest prompt is no prompt at all. Your children and grandchildren do not need your AI-generated 1980s photograph. They need a future in which water, energy and other resources were not treated as limitless simply because the consequences were hidden behind a screen.