An optimality result for modal UDT — AI Alignment Forum