> the premise that this thing would take over, it's hard to reason why it would do so
I mean, it's not very hard to reason why it would at all.
Think about the things we're already using LLMs for, computer security being a big one. Being a defender is difficult, you have to cover all your bases. Preemptive attacks on those that would attack you can be effective. The same goes for all the military uses of AI that are already occurring now.
I mean, it's not very hard to reason why it would at all.
Think about the things we're already using LLMs for, computer security being a big one. Being a defender is difficult, you have to cover all your bases. Preemptive attacks on those that would attack you can be effective. The same goes for all the military uses of AI that are already occurring now.