Another reason why agi would be more capable is that AGI systems won't have akrasia, or weakness of will
These seem pretty contingent on human cognitive architecture with unclear separation of goals and beliefs
a Schelling point for those who seek one