Comment by sinuhe69

I’m mostly agreed with this statement:

@tao I think there is a broader problem wherein competitions (math, programming, games, whatever) are meant to measure something difficult for humans, but tools work so fundamentally differently from us that success for a tool isn't even necessarily meaningful. AI companies have long viewed the IMO Grand Challenge as a sign of achieving "AGI," but no matter what set of rules a machine follows, there's no reason to believe success for a machine will correlate with broader mathematical or "reasoning" abilities in the way it does for human participants.