I would argue that IQ tests don't measure every single aspect of human intelligence, it's interested in one specific construct, if it exists -- g, which stands for a general capability. IQ is a test made for humans, but the concept of g is far more general than humans, and applies to machines as well.

When Google's AlphaGo beat some famous champions, people were wondering whether Google's program was merely a Go-specific program, or a more general program that learned to play Go.

I would also argue that if you went out and created another test for IQ, it would likely have high iter-rater reliability with IQ, but maybe have rough edges and have less empirical attention than mainstream tests like Ravens Matrices. That has been the history of people trying to make a name for themselves by making another intelligence construct, only to find it has insufficient distinction from IQ.