GIỚI THIỆU VỀ TẬP NÀY
What does it mean for AI systems to have "goals"? How could those goals differ from human goals? Why are some researchers calling for an "off-switch," and what would it look like? What are different future scenarios MIRI has mapped out and how could these play out on the international stage?
In this episode, we interview Aaron Scher from the Machine Intelligence Research Institute (MIRI). As an organization, MIRI has played an influential role in shaping the AI safety movement.
Aaron speaks in his personal capacity as a technical governance researcher. We go through a brief history of AI, the risks these systems could pose, the current policy landscape, and what can be done about it. We talk through scenarios MIRI has written about, as well as the work being done to try to mitigate them, and discuss the feasibility of international coordination on AI. We close out with Aaron's own path from a psychology degree into AI safety.
