Maybe transparent AI is the only way to fairly govern masses of people?

Let people regularly vote on which prompts should be added/removed, and have the AI justify all of its decisions, show which information it used etc.

I was thinking of discussion platforms but for general government I had a great idea: we know how to write and publish new laws but we are terrible at unpublishing outdated garbage laws. AI would be great for finding all laws that it thinks are dubious, impractical or otherwise undesirable in today's context and zeitgeist. There can be various manual deletion processes for different categories of absurdity.

An automated internet trial with robot judges and lawyers nominating users for warning banners of shame.

I believe this is likely true and will become evident in time to most, but not all, carbon based systems.

That's the most dystopian thing I've read today

Dystopia is the status quo: A handful of people controlling access for millions.