[Linkpost] “METR: Measuring AI Ability to Complete Long Tasks” by Zach Stein-Perlman

od LessWrong (Curated & Popular)