Nothing Went Foom: What Twenty Years Without an AI Apocalypse Does and Doesn’t Prove


Posted on Nov 21, 2026. All pictures shown are for illustration purpose only.
They said the first truly powerful artificial intelligence would wake up, rewrite itself, and race past human control in a weekend. The warning was not a joke. Serious people made a serious case, and it shaped research budgets, new institutions, and a whole culture of fear and urgency.
Twenty years later, the machines are better and faster. They write code, fold proteins, pass hard exams, and sometimes fail in public. They have not blown up the world, and they have not shown a sudden takeoff. Nothing Went Foom walks through the old alarms that faded, from famine forecasts and Y2K to models once called too dangerous to release. It checks famous ideas against systems that were actually built, then looks at recent agent incidents, the race to deploy, the fight over rules, and the price of slowing down.
The book gives optimists and pessimists their strongest arguments, then says which claims still hold. A missing catastrophe is evidence. The open question is evidence of what, and who pays if we read it wrong?