Lots of concern regarding super intelligent AI recently.
Roko Mijic warned about the risks years ago and thinks we should break up OpenAI.
Full episode -
Login to reply
Replies (7)
I think we should go faster.
You know,
fuck around and find out. 🫡
The Basilisk haunts me 😑
I actually didn't realise this was the Roko guy, I assume he named himself after the thought experiment lol 🤦♂️
Let me see if my English holds up here, but super and artificial is a paradox to intelligent, special case when a smartie pants illusion champion.
Wouldn’t wanna call you out for being an incompetent thief having hangers on notices.
View quoted note →
View quoted note →The Basilisk was a 2010 decision theory thought experiment posted on LessWrong, and many decision theorists dismiss it as fundamentally flawed.
I haven’t listened to such a lopsided take on AI like this in quite a while. Completely misrepresenting the Hugging Face incident as a “hack” is just silly! Anthropomorphising the “AI’s” seems weird as does not mentioning any sort of defensive alignment work. He seemed to be completely committed to the inevitable paperclip maximiser outcome. The whole thing was weird and almost unhinged.
I haven’t listened to such a lopsided take on AI like this in quite a while. Completely misrepresenting the Hugging Face incident as a “hack” is just silly! Anthropomorphising the “AI’s” seems weird as does not mentioning any sort of defensive alignment work. He seemed to be completely committed to the inevitable paperclip maximiser outcome. The whole thing was weird and almost unhinged!
View quoted note →