Comments by "" (@TheHuxleyAgnostic) on "Sam Harris on AI and GPT-4" video.

  1. "when manipulated" is the key part, because no AI sets its own primary subjective goals, and likely never will. The danger will always lie with the programmers, and their goals. No AI will do what absolutely nobody wants, like Sam suggested, because AI doesn't actually care, itself, about the outcomes. Someone has to program their desired outcome into the AI.
    1