Directly testing people's IQ through say contrived interview questions (ever wondered why people care about leetcode[0]?) will always be more accurate than pred

Directly testing people's IQ through say contrived interview questions (ever wondered why people care about leetcode[0]?) will always be more accurate than predicting it from their DNA. It is difficult to say how much of IQ is genetic, but estimates range from 50% to 80%, placing an upper bound on the effectiveness of the exact technique they used in GATTACA. (But, I want to emphasize, the more accurate way of doing it is already being done.) [0] One objection to calling leetcode an IQ test would be that some people "grind leetcode," as in, practice at it to get better, but did you know that Raven's progressive matrices has a large practice effect? In fact, I might raise your score just by telling you that a lot of the questions are based around XORing the presence of symbols in the pictures. If companies were giving culture-free IQ tests rather than leetcode ones we'd see the exact same pre-interview behaviors, including practicing tasks that don't resemble the job being interviewed for. [1] https://en.wikipedia.org/wiki/Heritability_of_IQ#Estimates https://en.wikipedia.org/wiki/Heritability_of_IQ#Estimates