Omnipotence test for AI safety - History — AI Alignment Forum