What is the RL environment to make an AI as good at politics as Lyndon Johnson, or as good at building a space launch business as Elon Musk? You can't easily containerize such domains, and so how would even a superintelligence trained in a datacenter be competitive against humans in such domains? @RyanGreenblatt thinks that transfer learning from further scaled up RLVR training is high enough that future AIs will be superhuman in their ability to learn how to operate in such domains…