However, it is only as dynamically inconsistent as the human.
Even if the inconsistency is on the human level, in cases where capability of AI is much higher, then that can cause problems on a larger scale than humans do. In other words, being as inconsistent as humans does not guarantee being as safe as humans.
Even if the inconsistency is on the human level, in cases where capability of AI is much higher, then that can cause problems on a larger scale than humans do. In other words, being as inconsistent as humans does not guarantee being as safe as humans.