It Looks Like You're Trying to Take over the World
lesswrong.com
lesswrong.com
Jokes aside, though, it's a really thought-provoking read, crammed full of references to actual research papers, giving a plausible-sounding step by step explanation, with numbers, of how an AI hard takeoff scenario would look.
Whether as a piece of fiction or as a call to take seriously how poor our defences are against such a scenario, I think it is well worth a read by anyone interested in AI. I remember a few years ago I was searching for a description of what an AI box escape might involve, so this was really satisfying to find.