Good point.

We trust AI outputs without checking the math behind them. Does the proof run for every supported query, or only for specific calculations?
$SXT