In Reply to: The difference between the two phenomena is almost infinite posted by TheHappyBurgermeister on September 14, 2026 at 11:49:26
(and ideally both): (a) you can confidently predict the chain of events your action will lead to, and/or (b) you need to have a way of containing the action in case it spirals out of control.
... which is why you don't do things like attacking Iran willy nilly, but I digress.
In the case of AI, we have no way to predict the chain of events they can start, and given that there are so many of them right now, whether out there in the wild or waiting to jailbreak simply increases the likelihood of something to spiral out of control in fully unpredictable ways. We also have no way of stopping a runaway event by any means other than shutting down all devices connected to the network and/or has even minor processing capability and purge the whole hardware all at once. Which ... I mean, imagine the logistics of that and the catastrophe that would bring.
Now just to make the problem even worse, consider the fact that these suckers move very very very fast such that by the time we decide we'll take some counter-action, these machines probably have already known about it 0.3 seconds into its existence and has developed, tested, and deployed its own countermeasures about an hour 15 minutes ago.
Yeah. Not very good, is it?
Ironically, aggressively pre-tech societies who don't even know what a cell phone is probably have a better chance of surviving what's coming.