Even if there is a limited number a A.I. in the gov., those A.I. could out perform humans on specific tasks easily. Examples are researching facts, bring up relevant information, hold persons accountable, and people could still make the important decisions A.I. couldn't.
Assuming it's true, an algorithmic gov't. capable of handling the complexity of human affairs "better" would be equivalent to strong A.I.; but our existing strong A.I.s (us) are demonstrably not capable of governing us as well as something algorithmic. Your premise is false by reductio ad absurdum.