My stats professor in 2019 said humans would always catch the bias, and that claim has aged terribly
She promised us a human reviewer would catch every skewed hiring model, but 3 years at a Chicago HR tech firm showed me reviewers rubber stamp 200+ flagged resumes a day, so who is actually supposed to be the backstop here?
Rubber stamping 200 resumes a day is bad, sure, but is that really proof the whole idea falls apart? A lot of those flagged ones are probably obvious calls anyway, like a model dinging someone for a gap year or a weird keyword match, so the reviewer is basically confirming what a quick skim already tells them. The real test is whether the truly bad ones slip through, and honestly, most hiring still comes down to a manager interview at the end. Your professor was wrong about humans catching everything, but I'm not sold that it's a total disaster either.