LW - Talk: "AI Would Be A Lot Less Alarming If We Understood Agents" by johnswentworth | The Nonlinear Library | Podwise