Again, this is an easy and thoughtless take without a proposed alternative. (Benevolent AI is just benevolent dictatorship with a veneer.)

It's also gratingly at odds with very modern history. History that, granted, is not part of our public education corpus.

A benevolent AI would not work as humans will be the ones to define the constraints under with the benevolent AI operates. Make no mistake, the ruling class will define constraints that will give them more power.

My remark about "humans at the helm" may have given you the false impression that I am optimistic about technology solving our societal woes. I apologize for that. I state that there's no eucatastrophe to save us from ourselves. My remark was intended as a sardonic statement that while there are humans, nothing will ever be improved. These problems are as old as humanity, and will only cease being problems when humanity ceases to exist.

Also it's pretty easy to list actual small tweaks. Does that count as improving democracy itself?

Education works, even if its "efficacy" is pretty low.

Simple solutions aren't the same as easy ones.

Education absolutely does work wonders on social and economic fronts, but it is unevenly distributed and the administration of it is easily subverted.