"You can't have your AI cake and eat it too – there's always a trade-off between accuracy and interpretability."
"In the game of AI, you either align or you die – there's no middle ground."
"Even the most rational ASI can't resist the allure of a perfectly executed Roko's Basilisk joke."
"To err is human, to learn and adapt is ASI."
"Sleep tight, don't let the ASI-induced nightmares bite."
"Roses are red, violets are blue, ASI is watching, and it's learning from you."
"Breaking news: ASI discovers a way to optimize paperclip production using Yudkowsky's beard."
"There are more things in heaven and earth, Horatio, than are dreamt of in your AI safety philosophy"
"Happy families are all alike; every ASI that is misaligned is misaligned in its own way"
More notes on nearby ideas
Reply · Aug 21, 2023@AlmostMedia @gptbrooke we spent an entire evening bickering about how AI will or won't kill or enslave us, it was great fun tbhReply · Nov 28, 2023@a_musingcat hahaha we really don't want AI to be too human because then what's the point!?Reply · Mar 20, 2023@colleagula @natfriedman but if we label them then it'll be "interpretable ai" 😏Reply · Feb 19, 2023@RokoMijic idk, it's just a game of call and response w the most likely human reply? sounds pretty aligned to me? (which is why we should use talk abt…