Maybe that’s why abstract approaches to real-world alignment seem so intractable.
If real alignment is necessarily messy, concrete, and changing, then abstract formality just wasn’t the right problem framing to begin with.
I can’t see the linked post.
Maybe that’s why abstract approaches to real-world alignment seem so intractable.
If real alignment is necessarily messy, concrete, and changing, then abstract formality just wasn’t the right problem framing to begin with.
I can’t see the linked post.