Sneaking in Connotations — AI Alignment Forum