AI is really *really* bad for the planet, here’s why

External

Let’s face it, AI is everywhere. Specifically, ‘generative’ AI models that produce the photos and videos we see online, down to the text answers we get from Google searches. Artificial Intelligence is here, and it’s here to stay. But is this a good thing? Not everyone has been keen to adopt AI into their daily lives and for a multitude of different reasons. The one I’m going to look at here is the impact of using AI on the planet.

‍

Firstly, it’s important to understand how generative artificial intelligence models work. In a very simplified manner, these models are incomprehensibly large collectives of algorithms and logic gates that process an input, and produce an output (text, pictures, audio, …) so convincing that the model is considered artificially intelligent. To copy an analogy by CGP Grey, think of AI models like a brain, while each neuron’s function can be understood, and groups of neurons vaguely grasped, the whole is a mystery^[1].

‍

These complex algorithms live on servers, which live in data centres. Of which are popping up more and more across countries like the US, the UK, and Germany^[2] to support the ever-growing number of companies who want to include AI in their products in some way. Data centres are not just part of the problem, they are the problem. Why? Water consumption.

‍

As Beverley Morris, a Georgian resident, explained to the BBC in July of this year, the sheer volume of water used by a local data centre to keep their servers cool has dried up the well she used for water, causing it to be low pressure and full of sediment^[2].

‍

“I’m afraid to drink the water, but I still cook with it, and brush my teeth with it. Am I worried about it? Yes.^[2]”

Beverley Morris, to the BBC

‍

Morris reported that even after having to pay a plumber to restore water pressure in her kitchen, it still contains an alarming amount of residue^[2].

‍

This is just the case for 1 resident near 1 of over 10 000 data centres globally.

‍

When a 1 megawatt data centre can use up to 25.5 million litres of water a year to cool their servers, equivalent to the daily water use of 300 000 people, it is clear to see the issue^[3].

‍

Water is the source of life on Earth, it is arguably the most precious resource we have on this planet, and we simply cannot rationalise the quantities of water that AI necessitates.

‍

Our water is not up for sale.

‍