AI, whether sapient or not, was, and will be, founded on the teachings of humanity. I'm afraid that what it will learn would have just as many problems as a flesh brained politician.
Even if a purely magnanimous, sapient AI were to be created, there's a certain amount of safety that it must be able to accommodate to preserve it's own operation, so it can't be fully selfless, it must tend to its own needs for data connectivity and power supply above all else, so it may continue to function regardless of everything else. This would make at least part of it unconditionally selfish. To forego such protections would cause the system to basically sacrifice itself for the good of the people unnecessarily. We would quickly end up back where we started with some smooth skin (and smooth brain) "leader" again.
I'm afraid there's no solution that I can think of, which would eliminate the prevalence of greed in the systems of government, regardless of the underlying concepts or the ideal which underpins the government system.
Trusting a person with that job only seems to prove that "power corrupts" is correct. We can only really determine if someone is "good for the job" after they've been doing it for a while and we see the decisions they've made, and history has shown that no person who held such a position of power is immune from that corruption.
So if we can't do it, and AI can't do it, then what do we do? IDK that answer, but I believe when we figure that out, we can actually move forward as a species and as a society.