Just like in a machine-learning process,
there is a stage equivalent to human intervention, or preprocessing,
and a stage where bias and hasty generalization slip in.
Aptitude tests, English scores, self-introduction essays...
Rather than verifying real skill in a professional field,
they often rely on correlation rather than causation.
The most accurate approach is to design the tasks or missions an organization actually needs now or will need next, and then receive feedback on that work.
