Depends on the test. There's been a move away from tests that require cultural or language specific knowledge and which instead rely on "which shape in the sequence comes next" type questions but they have their own set of problems in that they don't capture linguistic ability which is a part of intelligence, and are also useless for blind people. Earlier IQ tests required a lot of culturally specific knowledge, and were often quite up front about it because they considered the possession of such knowledge to be a marker of intelligence. Generally that viewpoint is out of fashion now so test makers try to come up with tests that measure "pure" Intelligence, whatever that means.

Culturally specific knowledge obviously correlates with an ability to learn by the very definition of the ability to learn.

I doubt this viewpoint is out of fashion. The problem is that it is hard to compare people from different environments if you use knowledge that depends on the environment...

So does this mean Einstein's IQ was subjective? Was he supposedly smart because he had encountered things on the test before? Would he score lower on one today?

Not in 1971 when the lawsuit was filed. Now they use picture-based tests because those are supposedly less dependent on learned knowledge. (Though I find that many Raven's progressive matrices-type questions are a lot easier if you know about symmetry groups and XOR.)

> Though I find that many Raven's progressive matrices-type questions are a lot easier if you know about symmetry groups and XOR.

Yes, at that point people who have already developed certain mindware have an advantage, but arguably that's the point of an IQ test. You're not testing the potential of a person to eventually be really smart, you're testing for the current problem-solving potential.

Well, it is problem, because the assumption is that you're measuring a proxy for g that supposedly doesn't change. So if it can change, by something as simple as learning some boolean algebra, the IQ test isn't measuring what it is assumed to be measuring.

That being said, you're right that it would still be useful for measuring problem solving potential, but that's explicitly not what IQ tests are supposed to measure.

??? who assumes g can't change?

It's generally accepted in psychometrics that general intelligence does not change in adulthood : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3950413/

I personally think it's not something that's really true, but it's how I learned it in university and seems to be the default position of experts in the field.

This is of course assuming no traumatic brain injuries or the use of psychoactive substances.

thanks for pointing out the year. I've updated my grandparent comment to include 1971

This is very wrong. Most complete IQ tests contain a lot more than just progressive matrices, because there is such a thing as verbal IQ that can't always be measured well by progressive matrices.