That makes sense but maybe we don't need that explanation. It could really be possible that the best-designed decision making robot would use confidence as a he

That makes sense but maybe we don't need that explanation. It could really be possible that the best-designed decision making robot would use confidence as a heuristic. It would fail in the same ways we fail, but it would probably perform worse in general if you removed that bias.