AI is software, not hardware; it only exists and operates in computers and cyberspace. That being the case, how can it destroy civilization as we know it today?
Here are three scenarios:
- ECONOMIC DISASTER: A large portion of the world’s population is dependent on the internet in one way or another. Our economic system and our infrastructure rely on it. Rogue AI agents could essentially take over the internet without any help from humans; we would have to adjust by returning to the world of, say, 1990. The costs of doing so would be immense.
- AI KILLS ALL OF MANKIND: To use one example here, AI can identify a way to create a bioweapon that would destroy humanity, but it can’t actually do it by itself, because it doesn’t have a corporeal existence. For this to happen, it would have to operate through collaborators (either robots tied to the internet or disgruntled humans) or be so intelligent that it could turn itself into matter. The last scenario here would make AI truly godlike.
- AI ENSLAVES HUMANITY: Same as #2–it would require collaborators or the ability to escape cyberspace and become corporeal.