Less Wrong mentoring thread — AI Alignment Forum