The case for learning from relatively weak AI misalignment
A commenter sees a potential upside in “relatively weak” AI misalignment: chances to learn rather than be taken by surprise.
TLDR
A commenter argues that seeing “lots of relatively weak misalignment” in AI offers opportunities to learn instead of being taken by surprise. They argue this means we are not in the worst alignment timeline they say Bostrom described.
The case for learning from relatively weak AI misalignment
A commenter sees a potential upside in “relatively weak” AI misalignment: chances to learn rather than be taken by surprise.
TLDR
A commenter argues that seeing “lots of relatively weak misalignment” in AI offers opportunities to learn instead of being taken by surprise. They argue this means we are not in the worst alignment timeline they say Bostrom described.