x
Entropic Alignment: Aligned AI as Truth-Seeking — LessWrong