I agree with your main point, but note that a gender-height correlation is effectively the basis of a linear regression, the most basic of all ML models.