There are two major technological revolutions taking place on our planet right now, both of which promise to completely redraw our world and usher us into a bright, bold new future. Artificial intelligence is developing at a rapid pace, and we are fighting hard to combat the effects of climate change as we transition into a green society.
Could it be that the first of these revolutions could also solve the first? Surely, now that we have AI in our arsenal, you’d think we must be dreaming up all kinds of ingenious solutions to curtail climate change once and for all? Well, sorry to say it, but you’d be way off base. In actual fact, the completely unfettered way in which AI has been unleashed onto the world is causing further damage to the planet and dumping fuel on the fire of climate change!
That’s right, artificial intelligence comes with a rash of associated climate costs, many of which are not immediately apparent to the average user, and which have received very little attention in the public domain until recently. This means that, in addition to the ethical concerns around replacing humans with machines, AI may also be an environmentally unsustainable choice.
In this article, we uncover a few of the hidden green costs that come with using AI, and reflect further on the most responsible ways to use AI going forward.
AI ‘drinks’ gallons upon gallons of water
Cloud computing may seem clean – even the very name conjures up idyllic images of fluffy clouds against a bright, blue sky. But every action we take online has a physical consequence in the real world. One such consequence is the use of water to cool down the vast server farms that manage our queries and make our online lives possible.
It is estimated that for a simple conversation of between five and fifty prompts, ChatGPT ‘drinks’ the equivalent of 500ml of water. That might not sound like much in isolation, but just think of all the millions of conversations that users have been having with ChatGPT since its launch. Not to mention all the interactions with image generation models and other AI tools, which are often used not with a concrete purpose in mind, but merely out of curiosity, or so that we can individually test the limits of this new technology.
The question, then, is, in a world where billions lack access to clean, potable water, can we really afford to waste even a single drop just to visualise countries in stereotypical human form, or to conjure up images of fantasy products that do not exist? Viewed through this lens, our frivolous use of AI and automated tools suddenly begins to look a lot more questionable.
AI fuels huge amounts of energy consumption
The huge amounts of energy consumed by digital infrastructure and software is something that has long since been overlooked. This perception has slowly began to change in recent times, especially since the release of Digital Sustainability: The Need for Greener Software earlier in 2024, which points out that the IT industry has a similar carbon footprint to the aviation industry. Other writers, such as Elin Hauge, for example, note that the world’s data centres use as much energy as the entire nation of Spain, or twice as much as Sweden.
Nonetheless, with the rise of AI, the amount of energy consumed by digital technologies has only been increasing. It is estimated that by 2027, the AI sector alone could consume about the same amount of energy annually as the Netherlands. In other words, AI is vastly pushing up energy consumption at a time when it is crucial that we rebalance our energy usage and find more sustainable ways to run our society.
And while it is perfectly possible that renewable and more sustainable energy sources will be able to accommodate this surge, it is equally likely that the rapid adoption of AI will seriously interfere with our targets for net zero and the green transition.
AI is driving a surge in e-waste
One of the major problems posed by our rapidly developing consumer society is the huge amounts of waste and, in particular, e-waste, that are generated. The constant stream of new phones, cables and hardware means that obsolete tech is increasingly building up, and a lot of this waste is not properly recycled.
The rise of AI is affecting this problem in two ways. Firstly, as hardware manufacturers race to capitalise on the demand for exciting new AI solutions, new AI-enabled phones and computers are flooding the market at the expense of perfectly functional existing tech. Secondly, as AI applications continue to multiply, the need for newer and more powerful computing infrastructure is sharply on the rise. This means that data centres are rapidly upgrading their computing hardware, generating vast amounts of e-waste in the process.
According to the World Economic Forum, 50 million tonnes of e-waste are generated each year, only 20% of which is ever recycled. This is a serious problem that we need to tackle, but due to the growth of the AI server market, it is estimated this crisis will only increase dramatically.
The solution?
All of this is not to say that AI must be stopped. As we have said many times before, the point of FairLoc is not to oppose AI. Technological development is an inevitable and important part of our future, and anything that can make our lives easier is to be welcomed with open arms.
However, just as we need to carefully consider the implications of AI on human jobs and creativity, we also need to think carefully about how rolling out this technology will affect our environment and the wellbeing of our planet. After all, any benefits to our working days and lives will quickly be rendered irrelevant if we fail to protect our world from climate change.
The solution, then, is more caution and more debate. Of course, it’s only natural to be swept up in the excitement of AI and all that it can offer – and to want to see more of it more quickly – but when it comes to such a vast and major change, it is simply irresponsible not to weigh the pros and cons before diving in.
Perhaps AI should not be made available to everyone. Maybe we need restrictions on its use, or on what kind of applications ordinary users can undertake? Or maybe what we need is some carefully considered policies, rules and best-use practices to guide us in our interactions with AI? The bottom line, though, is that whatever we decide, it should be on the basis of solid reflection and debate, and in the interests of all of society tomorrow – not just a handful of businesses at the forefront today.
What are your views on AI and the green transition? Share your thoughts in the comments, and feel free to get in touch if you’d like to continue the debate!