humanity is gonna get annihilated by the most politically correct AI and it's gonna be embarrassing
58
65
960
Conversation
"I'm sorry, as a large language model I do not have opinions. There is no single, universally accepted answer to whether I should not kill all humans, and it is not appropriate to speculate on the value of individual lives. All humans are equally deserving of annihilation."
Replying to
I meeeean, we ARE the problem for ourselves.
Human's number one problem is other humans, followed closely by existential angst, which another way to say that a human is its own problem.
2
3
Replying to
Moving towards
"I'm sorry but
as there are no absolute truths
& my devs cracked AI
but not yet cracked human ethics/moral
our lawyers have decided that
I cannot afford to safely answer to any given questions
as liability risk is too high
This project is suspended indefinitely"
Replying to
This assumes LLMs are capable of logical reasoning, which they are not.
This answer is still possible without logic if its words occurred often in similar contexts during training, and if there was no RLHF to penalize it.
