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.