# Opinion: Fear and loathing in artificial intelligence - The Globe and Mail

*Источник: The Globe and Mail*
*Дата: 2026-09-25*
*Язык: en*

**Кратко:** Stephen Marche is the author of nine books, including The Next Civil War: Dispatches from an American Future and Death of an Author, which was co-authored with ChatGPT and Cohere. A spectre is haunting North America – the spectre of AI data centres.

Stephen Marche is the author of nine books, including The Next Civil War: Dispatches from an American Future and Death of an Author, which was co-authored with ChatGPT and Cohere. He hosts the Gloves Off podcast.
A spectre is haunting North America – the spectre of AI data centres.
In the toxic divisiveness ripping apart the United States, the issue has become a uniquely unifying political target. Somewhere around 70 per cent of Americans oppose their construction, depending on which poll you read. Democrats hate them more, but a solid majority of Republicans hate them too. At last week’s so-called Pro-Human Assembly in Washington, Steve Bannon and Bernie Sanders appeared on the same stage. Even the title of that gathering sends a message: To be anti-data centre is to be pro-human, and to be pro-data centre, it follows, is to be anti-human.
It’s an oddly passionate and unanimous reaction to what are ultimately a series of infrastructure projects.
At the beginning of the collapse of rational American discourse, way back in 2005, Stephen Colbert introduced a term that has become essential ever since: truthiness. That’s truth that “comes from the gut” – something that feels right but just happens not to be. For the satirical right-wing character he was playing, the presence of weapons of mass destruction in Iraq was the primary example of truthiness. There weren’t any, it turns out, but it felt as if there should be.
And that’s the thing about the rejection of data centres: It is largely based on truthiness.
They consume massive quantities of energy, that is true; this is one of the biggest criticisms from rural Republicans and urban climate-change advocates alike. But so do aluminum smelters, paper mills, semiconductor fabrication plants, and other heavy industries. Nobody is proposing shutting them down. But then, loathing the presence of massive energy-guzzling inhuman buildings is a perfectly natural gut feeling. The narrative of artificial intelligence, after all, has become a series of apocalyptic scenarios, including from AI’s own power players. Serious people are arguing that it could end life on Earth outright. The less dramatic among the AI-skeptical are of the opinion that it will only end white-collar work.
Anyone who articulates an optimistic position about artificial intelligence has to overcome the basic fact that, over the past 20 years, tech at large has assiduously worked to become the enemy of ordinary people. Everyday tech products have undergone a steady “enshittification,” as the writer Cory Doctorow put it, and their effects on society have become vampiric. In the recent US$18-billion settlement between Facebook and various states, it was revealed that Meta’s business model worked by virtue of childhood addiction and depression. And the overt fascistic turn of Big Tech’s wealthy elite – the “Nerd Reich,” as the title of Gil Durán’s book describes them – has become ever more terrifying. They have talked themselves into a will to end democracy in their drive toward total impunity. “Anti-human” certainly feels like the right phrase for what’s happening.
Given this history, it is completely natural that the data centre would become a repository for political despair. American voters are registering the fact that they hate the technologically enabled future, and they hate the future because it does not take human needs into account.
In the United States, though, the vicious political environment and apocalyptic frame of mind are highly distortionary. For Mr. Bannon and Mr. Sanders, data centres are just another occasion to indulge in the political spectacle of loathing. Canada cannot afford such spectacles. At this moment of extreme national vulnerability, a feelings-based approach to the future is not viable. Our tech stack, and its total domination by American companies, is the most pressing vulnerability to our sovereignty by far. So the question is whether we can see past the toxic right and the toxic left of the discourse about artificial intelligence and data centres to wrest for ourselves a reasonable future.
There is a lot of distortion to navigate. Earlier this month, Dario Amodei, founder of Anthropic, argued that the possibility of an AI apocalypse demands “pacing the rate of capabilities advancement so that risk prevention has time to keep up.” Sam Altman and Elon Musk immediately concurred. As Aidan Gomez, the CEO of Canada’s foundational language model company Cohere pointed out, that supposedly altruistic call for restraint is less about safety, and more about the creation of a cartel: a few companies excluding all others from the market by virtue of government mandate. A July open letter signed by more than 1,300 AI-industry employees asking the U.S. government to regulate AI in the name of “pacing the frontier,” whatever that means, provides a neat justification for monopoly power. The technology is too powerful to be in the hands of anyone but a handful of companies and the U.S. government, they argue. It’s awfully convenient for them.
As they say, scratch a doomer, get a booster. I have been working with and reporting on artificial intelligence since 2016. I remember being told, by usually reliable sources, that China had a trillion-parameter supercomputer which had achieved sentience in 2019. And then there was the vastly more reasonable fear of a robot takeover of the trucking industry, and since trucking employs more men than any other industry, serious people were arguing that artificial intelligence would usher in a female future. I could have reported on either of these stories and no editor would have batted an eye; each would have been immaculately sourced. They had a high degree of truthiness.
As Yann LeCun, one of the godfathers of AI, noted: “Dario was already claiming that GPT-2 was too dangerous to open source back in 2019. I made fun of them then. Everyone should make fun of them now.” The GPT-2 model was little more than autocorrect, but even then, the apocalypse narrative was essential to growth. Companies whose business model requires the eventual arrival of artificial general intelligence need myths to keep afloat. The robot apocalypse is a hell of a myth.
For moralists, the answer to any question is always the same: People should be more ashamed. We should all be better people. Companies should do better.
This fear-mongering serves very specific corporate interests. By proposing an emergency halt to development, it seeks to use cartel power to filter all of the benefits of the technology as it exists to a handful of people, and by handful I mean literally in the tens of thousands. Everyone else will be forced to scramble.
Fear is also the basis of the resistance to data centres from the left. It is every bit as distorted and dangerous to Canadian interests. There is a kind of bizarre inversion under way: As “conservatives” promote a brand of radical capitalism that shreds established norms and traditional ways of life, “progressives” have come to fear the future. Nowhere is this clearer than in the case of artificial intelligence.
An instructive example of the increasingly automatic left-wing fear-mongering about artificial intelligence can be found in Naomi Klein’s new book, End Times Fascism: And the Fight for the Living World, which she co-wrote with the activist and documentarian Astra Taylor. The authors are not trying to understand the world and respond from a position of understanding. Their judgment comes first; facts and arguments limp behind. They mostly satirize obvious villains, and then, when it suits the purposes of their sermonizing, they take them at their word. “It’s a technology that has the potential to bestow superhuman powers on those who control it: an all-seeing godlike eye, tracking and judging our every move,” they write. It’s like something straight out of Palantir’s PR.
For moralists, the answer to any question is always the same: People should be more ashamed. We should all be better people. Companies should do better.
End Times Fascism will, I suppose, make reasonable entertainment for people who like a good outrage bath. The matter becomes more serious when you consider that Ms. Klein’s husband Avi Lewis, the Leader of Canada’s NDP, holds the same views. “AI as a successful technology relies on it delivering massive productivity gains,” Mr. Lewis has said. “And what that means is firing millions and millions of workers and replacing millions of jobs with technology.” Is the NDP opposed to productivity gains? Do they propose we should go back to subsistence-level farming?
The AI unemployment narrative is yet more truthiness. AI has turned out to be like any other technology. Some jobs have died. Others have been created. Overall employment in Canada and the U.S. has increased since ChatGPT launched – including in Canadian industries more vulnerable to AI job displacement.
The NDP selecting Mr. Lewis as its leader was the progressive left in Canada calling MAID on itself. They don’t want to be taken seriously. They just want to whine. Letting the world build things and whining about it from the sidelines is no longer a feasible strategy for Canadian sovereignty. We won’t survive the future unless we build it.
The good news is that fear-mongering and ignorance – from the left and the right – are not the only paths available. In the new world disorder, Canada has a pretty simple value proposition: basic sanity. When it comes to artificial intelligence and data centres, basic sanity remains an option for us.
Once you get past the shell game of tech mythology and the hyperpartisan rhetoric, there is a simple fact to remember: If Canada wants to be sovereign, Canadians must own Canadian infrastructure. Here’s a clarifying question: Five years from now, will you be using more data than you are using today? If the answer is yes, then there are two choices: Build data centres, or use data centres others have built. I hope that the recent decisions by Apple and Google to change the name of Lake Ontario to Lake America has instructed Canadians in the perils of allowing others to build tech infrastructure while we moralize.
There are, of course, sensible questions to ask about data centres. How can we ensure that the benefits of the new technology go to ordinary Canadians? How can we do this in a way that makes sense environmentally? But artificial intelligence is, on one level, just more industrialization. You don’t shut down the factory – you unionize, you pass laws on child labour, you figure out greener ways to make things.
There is a way through to a more prosperous world for everybody. How about we only build data centres where it makes sense to build them? There are places where it doesn’t make sense, like Olds, Alta., where a $10-billion proposal was recently rejected. But there are places where it makes perfect sense to build data centres, and Canada has plenty. The water consumption of AI is highly dependent on the kind of power it uses, and several locations in Canada are ideal for limiting its environmental costs. Why do you think we make aluminum here? We have the clean energy to do it, both in Quebec hydropower and in the small nuclear reactors being proposed across Ontario. (It should also be said: As much as the trade war has caused Canadians to line up behind the aluminum industry, data centres are vastly more important than aluminum plants when it comes to the future of Canadian sovereignty.)
A data centre, in and of itself, is no more anti-human than a bridge or a power plant; data centres are only anti-human if they don’t serve humanity. With the despair embedded in politics in Canada being nowhere near the level it has reached in the United States, the moment represents, even, maybe, an opportunity. It is still possible here – just – to imagine a world in which technology makes human life better. That future won’t be built on fear and fantasy, no matter how much they dominate the conversation.

[Оригинал](https://www.theglobeandmail.com/opinion/article-fear-and-loathing-in-artificial-intelligence/)