“Any sufficiently advanced technology,” wrote the science fiction writer and futurist Arthur C Clarke, “is indistinguishable from magic”. This sentence — which Clarke himself somewhat grandly billed as Clarke’s Third Law — feels intuitively, and surprisingly, true.
I have some sense of how an old landline telephone works, vague though that understanding might be: the sound of my voice is converted into electronic signals that are transmitted, by way of cables and wires, to the receiver in the hand of a distant interlocutor, where it is converted back into sound. The internal combustion engine in a petrol car works by producing heat that generates sufficient energy to drive a system of pistons. Such understandings may be vague, but the point is that I have some working notion of what goes on in these kinds of analogue technologies.
I have no idea what on earth is going on in there, in the dark, occult interior of that strange machine. It feels, in the Arthur C Clarke sense, like a kind of magic
But if you asked me how ChatGPT can take a prompt — like, say: write me a 1,000-word Irish Times column about the environmental impacts of artificial intelligence in the digressive and pretentiously leftist style of Mark O’Connell — and instantly deliver a more-or-less serviceable text in response, I would be at a total loss. (This is a purely hypothetical example, by the way: the current iteration of ChatGPT is incapable of reproducing the specific mode of pretentious leftism I’m after, and so I remain in the position of having to laboriously think up and type out this stuff myself — at least until the next update drops.) I have no idea what on earth is going on in there, in the dark, occult interior of that strange machine. It feels, in the Arthur C Clarke sense, like a kind of magic.
Artificial intelligence, in general, has about it a sort of airiness, as though its effects were conjured from the ether, pure and insubstantial. And this is true, more broadly, of almost all information technology. It’s hard to conceive of any of the magical capacities that are contained within my iPhone — the ability to play practically any song that has ever been recorded, for instance, or to receive, wherever I happen to be, a live video feed of the goings-on at the front door of my house — as being tethered to real, physical technologies, requiring energy, dependent on fuel extracted from the ground.
Women don’t want a rest on Nollaig na mBan. We want decent childcare, equal pay and access to healthcare
What U2 knew about the United States that some quintessentially English stars didn’t
Crises and controversy: our predictions on what’s next for education in 2025
Fiction in translation: Alejandro Zambra rescues fatherhood from the box-ticking dutifulness of parenting manuals
Consider the term that contains, in its vast and vaporous embrace, so much of how we think (and fail to think) about our current technologies: “the Cloud”. It implies a great floating mass, as light and clean and insubstantial as the air itself. In her brilliant book Atlas of AI, the scholar Kate Crawford — one of a handful of indispensable contemporary thinkers about our technological moment — writes about the myth of clean tech and how it is sustained by such language. “The tech sector,” she writes, “heavily publicises its environmental policies, sustainability initiatives, and plans to address climate-related problems using AI as a problem-solving tool. It is all part of a highly produced public image of a sustainable tech industry with no carbon emissions. In reality, it takes a gargantuan amount of energy to run the computational infrastructure of Amazon Web Services or Microsoft’s Azure, and the carbon footprint of the AI systems that run on those platforms is growing.”
Entire sectors of the employment economy may be entirely transformed or even wiped out
It has, in recent years, become increasingly obvious that the advent of artificial intelligence marks not just a technological but a societal inflexion point. The economic and political upheavals it will bring — and is, in fact, already bringing — may be even greater than the scientific and technological innovations it promises. Entire sectors of the employment economy may be entirely transformed or even wiped out. It is already beginning to change how we conceive of such nebulous (which is to say, cloud-like) concepts as knowledge, intelligence, and creativity. But this inflexion point is a double one because we reach it at a time when we are already on the cusp of another civilisational turning point, around the fulcrum of a climate crisis. We find ourselves, in this sense, at a dizzying historical impasse: a paradigm shift taking place on a terrain that is itself undergoing an even more profound tectonic displacement.
And these shifts, these movements of crisis and opportunity, are deeply entangled with one another. Even as we are asked to believe in the potential for transformative technology to generate solutions to the climate crisis, we are forced to reckon with the sheer scale of AI’s demands on the environment. One infinitesimal, and therefore startling, example: the requirements for cooling AI server towers are such that for every 100 words of text generated by ChatGPT, three litres of water are consumed. The vast server farms and data centres that facilitate this technology are extraordinarily energy-intensive. The demands on the electrical grid, and the impact on the environment and the climate, are immense. And they are only getting started.
A huge amount of wealth is, of course, generated by, or channelled through, these data centres; in the economic style to which we have become accustomed, however, very little of it winds up in the pockets of people living in this country
Within this system, our country plays a crucial and disproportionate role. In recent decades, an exceptionally dense network of data centres and server farms has been built around the Dublin area. “The largest industrial development rollout this country has ever seen in terms of pace, scale and impact on our energy infrastructure,” as Una Mullally put it, in these pages, last November. An astonishing 21 per cent of our electricity is now consumed by these data centres; they are on track, by the end of this decade, to consume one-third — at a time when the majority of our electricity is still drawn from fossil fuels like gas, coal, peat and oil. A huge amount of wealth is, of course, generated by, or channelled through, these data centres; in the economic style to which we have become accustomed, however, very little of it winds up in the pockets of people living in this country. Only about 16,000 people work in these places. As the program of AI automation proceeds, we can expect that paltry number to reduce.
Despite the grand claims made by its developers and owners, the benefit of artificial intelligence to humanity, in general, is hard to measure or predict and may turn out to be negligible. It is, however, extraordinarily profitable for those who control it. The extraction from the earth of fossil energy; the extraction from human beings of knowledge and labour; the automation of that knowledge and labour by technology: this is the logic of artificial intelligence. It has also always been the logic of capitalism. There is, in the end, nothing magical, nothing airy, about this technology, or the economic system it grows out of and intensifies. It’s a machine of exploitation and despoliation all the way down: fossil energy burning, pistons pumping, wealth extracted and funnelled ceaselessly upward, to the owners of the machines, floating up there above us in their cloud.