My overarching thought after reading as much as I could stand of this is: "Wow, sure sounds like Zuck really drank his own Kool-Aid - or if he didn't, at least he wants us all to believe that he did!". The question is not whether "superintelligence" will be accomplished, the question is only how fast. I mean, LLMs are nice, but the jury's still out on whether they alone will be enough for "superintelligence"... > I d…
AI will almost certainly get to be pretty intelligent. Whether the algo is LLM or something else doesn't matter that much. I note the term LLM doesn't appear in Zuck's thing.