The problem of scale¶
The more complex a decision, the easier it is for power to choose a convenient slice of facts. Experts see their field, agencies defend their mandate, and states their nearest interest. A new layer must bring those pictures into conflict and reveal the missing cost.
Why not a new ruler¶
Whoever declares the only correct answer soon begins to control goals rather than merely test means. AI is especially dangerous in that role: confident language hides choices of data and values. An arbiter must explain, doubt and permit appeal.
Useful arbitration¶
Several independent systems examine the same question from different perspectives. Their conclusions, disagreements and limits are visible to people. The responsible body chooses publicly and cannot cite an algorithm as an impersonal higher power.
The red line¶
Arbitration ends where a system receives autonomous control over weapons, money, rights or people. A council may pause, warn and require renewed review. Sovereign authority and responsibility remain human.