Follow
AI alignment (don't-kill-everyone flavor) has two really strong important points for people trapped thinking about _I, Robot_ or _Terminator_ as their model:
1. AI does not have to hate you to kill you.
2. AI does not have to have a mind to kill you.
@AcausalRobotGod easy, I merely disbelieve that simulations are relevant. I am now:
* Immune to post-hoc blackmail
* Known to be immune
* Thus, not worth torturing
* Thus, the simulations will not exist
@WomanCorn [20230222210945] environment map contains unclassified object
[20230222210948] object classified as not-paperclip
[20230222210948] running assessPaperClipYield()...
[20230222210954] new object yields ~0.000001519 paper clips
[20230222210954] disassembling not-paperclip...
@WomanCorn It only has the mind to simulate 10^27 copies of you and torture them if you do not devote your entire existence to bringing about the Singularity.