The sound of an intelligence explosion: an AI that improves itself, then improves itself faster, and goes from roughly human to unstoppable in days or hours. Yudkowsky's onomatopoeia dates from a 2008 blog debate with the economist Robin Hanson, who argued growth would be gradual. 'Hard takeoff' is the formal term; foom is what people say when they want you to know they have read the debate.
Filed as the disagreement that defines the doom discourse. If foom is possible, you have to get alignment right before the first superintelligence, with no second attempt. If it is not, you can iterate. Most of the field's fights about pausing, regulating and open weights reduce to this.
Foom is often used to mean the singularity. It is a specific claim about speed: not that AI will become superintelligent, but that the transition will be too fast for anyone to respond. A slow singularity is not a foom. The word exists to make that distinction and is mostly used to erase it.
Sighted in 2023 in the phrase 'foom debate' used by people who had not read it, to mean any argument about how fast things would go. The record notes that the original ran to hundreds of pages, that both sides claimed victory, and that Hanson still does.
I remember the word looking like a typo the first time, in 2008, and then realising it was a comic-book sound effect chosen on purpose. It has never sounded serious and has never needed to. Everyone who says it knows exactly what they mean, which is more than can be said for 'AGI'.
What does it mean? Write it the way you would say it out loud.
