Suppose, our mind is indeed designed to have a postdictive illusion of choice, does it make sense to equip an AI with it? Does an AI need to feel in control? W

Suppose, our mind is indeed designed to have a postdictive illusion of choice, does it make sense to equip an AI with it? Does an AI need to feel in control? What kind of role can an illusion of choice have in reinforcement learning? Wouldn't it introduce errors on smaller time scales, e.g. when catching a ball? Or in building up procedural memory in general?