AI incompetence often comes from misalignment (opens in new tab)

Sometimes I see people say “I’m not worried about AI risk because AIs are really bad at things.” I think this is a misunderstanding.