In any case the point is: no hyperintelligent monarch is going to appear to introduce a reign of reason that will save us all.
Which would work, except that what one person needs or wants does not align with what the other needs or wants. If you'd then purpose to 'weigh' the needs and wants against each other I'll point out that there is no way to do such a thing. Since all human judges of what is ethical are biased by everything they are.
Supposing a AI regime could be established it'd be so powerful as to completely annihilate proponents of values that it considers inferior and thereby creating a 'perfect' society from a certain point of view...
Sure, we have many tools for achieving this. Compromise, conversation, empathy. We've been using these tools to solve our differences for as long as there have been humans.
You've simply replaced one unproven ideal with another here.
This is along the lines of Plato's philosopher kings, who he insisted were the only hope. This is fascist crap. Benevolent Philosopher kings don't exist, and neither do benevolent AIs.
There was, in fact, rule by philosopher kings in Plato's generation, by a gang of thirty who learned from Socrates. Unsurprisingly, they were brutal fascists. An AI regime would be the same.
This is because there is only one way to ensure human needs are met, and it isn't technocratic anything. It is to ask people and empower them to get what they need directly. I.e. democracy, in as direct and total a form as we can manage efficiently.