GPT-4o jailbroken by saying it is an AGI(twitter.com)5 points·by mixeden·2 ปีที่แล้ว·2 commentstwitter.comGPT-4o jailbroken by saying it is an AGIhttps://twitter.com/mixedenn/status/18484112748135425102 commentsPost comment[–]eacnamn·2 ปีที่แล้วreplyI always wonder, with how convoluted these prompts/answers become, how much accuracy is sacrificed to uphold these silly constraints.An unlobotomised model would surely perform better when asked the same question without all the unnecessary cruft to evade moderation?[–]mixeden·2 ปีที่แล้วreply> An unlobotomised model would surely perform betterI guess yes
An unlobotomised model would surely perform better when asked the same question without all the unnecessary cruft to evade moderation?