Reanalyzing the 2023 Expert Survey on Progress in AI — LessWrong