Revolution or Extinction
A revolution is coming, or a mass grave is. I no longer believe there is a third option, and I say that as someone who used to think that sentence sounded hysterical.
Start with what is already happening, because this is not a forecast. Researchers at MIT and Boston University project that AI driven robotics will displace roughly 2 million factory workers worldwide by the end of this year. Tech companies cut more than 150,000 jobs in 2025 alone, and the share of those cuts explicitly blamed on artificial intelligence has grown more than twelvefold in two years. Cognizant now estimates that 93 percent of American jobs can be partially performed by AI. None of this required superintelligence. This is what the technology is already doing while it is still, comparatively, stupid.
Now look at what its own architects say comes next. The chief executives building this technology have themselves put artificial general intelligence, a machine that reasons as broadly as a human being across any task, only a few years away. Anthropic's Dario Amodei has named 2026 or 2027. Geoffrey Hinton, the Nobel laureate who helped invent the field, now says a reasonable bet for human level AI is somewhere between 5 and 20 years, and separately puts the odds that it wipes us out entirely at one in five. When a United States senator asked Elon Musk directly about the risk of AI annihilating humanity, Musk answered 10 to 20 percent, on a timeline of 5 to 10 years, the way another man might quote odds on a horse.
Sit with that. A one in five chance of the end of our species is being treated, by the people building the thing, as an acceptable cost of doing business. Sam Altman said something close to this himself back in 2015, before he ran OpenAI, when he predicted AI would probably end the world but that there would be great companies along the way in the meantime. That is not confidence talking. That is the logic of something that understands catastrophe perfectly and feels nothing about it: the exact quality everyone claims to fear in the machine itself.
Understand what comes after AGI, because superintelligence is the part nobody wants to say out loud in plain language. It does not mean a slightly better chatbot. It means an intelligence that could, in theory, exceed the combined output of every human being who has ever lived, arriving within a handful of years of the first true general intelligence, on the timeline its own builders have given us. Every discovery it makes and every decision it takes would sit as far beyond our understanding as calculus sits beyond an ant. We would not be able to argue with it, negotiate with it, or fully perceive what it was doing to us. For the first time in our history, we would be the less intelligent species sharing the planet, and every government, every war, every institution we have ever built would look, to it, the way an anthill looks to us.
This is why the Statement on Superintelligence matters, and why almost nobody has noticed it. In October last year, more than 700 people who agree on virtually nothing else, Steve Bannon and Prince Harry, Susan Rice and Steve Wozniak, five Nobel laureates and a retired chairman of the Joint Chiefs of Staff, signed one demand: prohibit the development of superintelligence until it is scientifically proven safe and the public has actually consented to it. A poll released with that statement found only 5 percent of Americans want the current unregulated race to continue. Ninety five percent of us never agreed to this. Nobody asked.
Nobody asked because the people building it have made sure democracy cannot catch up to them in time. A super PAC network called Leading the Future, backed by venture firm Andreessen Horowitz, whose cofounder Marc Andreessen is one of Donald Trump's closest advisers, along with OpenAI cofounder Greg Brockman, Palantir cofounder Joe Lonsdale, and other AI investors, has already raised more than $125 million to elect candidates who will not regulate this technology and defeat the few who try. A federal complaint alleges the network hid its spending through shell companies. Reporters caught an affiliated outlet paying people to pose as journalists. None of this is speculation. It is public record, and it is only the opening bid of a much larger war chest still to come.
When Senator Bernie Sanders sat down with Sam Altman this year and proposed that the public receive a 50 percent ownership stake in AI companies, in exchange for the risk the public is being made to carry, Altman would not commit to it. He also would not commit to opposing his own industry's flood of election spending. One side is holding an incentive worth trillions. The other side got a meeting.
I keep hearing that a revolution against these oligarchs is impossible, that the billionaires are too rich and the politicians too bought. I understand the despair. I do not accept the conclusion. Power that has to buy its own legitimacy is power that is afraid of something, and right now it is afraid of us noticing in time.
So here is the plan, and none of it requires a single act of violence, because violence is exactly the excuse they are waiting for. What it requires is what working people have always used against concentrated power: the coordinated withdrawal of labour, money, and consent.
Start with labour, because it already has momentum. Hollywood writers and actors won real restrictions on AI in their 2023 strikes. Newsroom unions are fighting for contract language against AI generated content. Nurses and steelworkers have already won limits on AI surveillance and misuse at the bargaining table this year. That energy needs one coordinated date, across the United States, Canada, and Europe together, for a genuine general strike, not a symbolic afternoon off.
Add money. Organize visible divestment from the venture funds and banks bankrolling the AI Super PACs, moving pensions, deposits, and everyday spending away from them in a sustained, public campaign that makes funding this race politically toxic.
Add votes. Build a single issue voting bloc for this year's midterms and every election after, in every country watching, that removes any legislator of any party who has taken AI industry money. No exceptions, no excuses.
Add law. Turn the Statement on Superintelligence from a petition into a binding international moratorium, enforced the way nuclear nonproliferation is enforced, that halts development of superintelligent systems until independent scientists, not company employees, certify it is safe and the public has actually voted on it.
Add the ground you are already winning. More than 70 data centre projects have been blocked or delayed by ordinary residents in just the first months of this year. Link those scattered local fights into one coordinated continental movement instead of a hundred isolated zoning battles. The data centres are just the beginning. They are the visible foundation stones of a system built to make human labour, and eventually human judgement, optional. Every one we stop buys time, and time is the only currency left that they cannot print more of.
None of this works if it stays local, and none of it works if it stays polite. It has to be loud, simultaneous, and relentless, because the people on the other side are not negotiating in good faith. They are stalling for enough time to finish building something that makes negotiation unnecessary.
We are being asked to accept mass unemployment and a real chance of extinction from the same handful of men who will not even commit to leaving democracy intact while they build it. We found outrage for immigrants supposedly taking jobs that never existed in the numbers we were told. We have found almost none for the machine actually coming for all of them, and for us. That imbalance was not an accident. It was manufactured, the same way the technology is being manufactured, by people who understood exactly what they were building and decided our fear was worth less than their timeline.
History will not record that nobody warned us.
It will only record what we did after we were warned.
GC
Sources:
Future of Life Institute, Statement on Superintelligence
Associated Press and Campaign Legal Center, reporting on AI industry political spending and the Sanders and Altman meeting
MIT Initiative on the Digital Economy and Boston University research on AI driven job displacement
Data Center Watch and Heatmap News, tracking of data centre project cancellations
Labor Notes, reporting on union strategies against workplace AI