x
Debate (AI safety technique) - History — LessWrong