Can AI be trusted with inclusion?