Obviously we can’t dig down into definitions forever. I’m content to let “goals” be vague. I’m not happy with leaving “intelligence” vague, because it’s abused so much.
To enshrine one goal (or set of goals) as the “right” ones to define intelligence, would be to defeat the purpose entirely. That’s exactly what I’m trying to avoid. A system which we give a specific goal, for instance to drive a car, we can talk about how “intelligent” it is with respect to that goal. But the intelligence of “AGI” we can’t talk about until we get more concrete about what we actually want from it.