Comment by rogerkirkness

It seems like Anthropic is increasingly confused that these non deterministic magic 8 balls are actually intelligent entities.

The biggest enemy of AI safety may end up being deeply confused AI safety researchers...