Over a population, outside factors affecting the score is smoothed out to create a normal distribution. Over one individual, most factors remain the same.

Over two different people, so many factors affect the score that making the claim "one person is more intelligent than the other" is statistically unsound without a massive score difference. This is even ignoring that a full IQ test involves FAR more than the usual online logic puzzles people tend to know, yet still have these flaws.

so yes; > colloquial use and understanding of IQ is flawed.