Intelligence doesn't even have a well-agreed definition, so of course it's going to be difficult to quantize it accurately. If I had to venture a guess, I'd say it's more a vector than a scalar.
Still, whatever it is IQ tests measure, there's significant evidence that they measure something, and that something has some correlation with intelligence. A test doesn't have to be perfect to be useful, if given the proper weight.
To paraphrase a quote, the source of which I've forgotten: if you can't measure what really matters, you measure what you can and hope it turns out to be relevant.