Loading…
16
Caught my own bias in a hiring model I built, felt sick
I spent 6 months fine-tuning a resume screener for a retail chain in Cleveland. Last Tuesday I ran a test on old hires and saw it was docking points for gaps in employment, no matter the reason. Then I realized I had copied that rule from a template I never questioned, and it was hurting single parents and people who took care of sick relatives. It took a friend asking 'why is that even a factor?' for me to see I was the problem, not just the code. Has anyone else had that gut-drop moment where you find a bias you coded by accident and had to redo the whole approach?
1 comments
Log in to join the discussion
Log In1 Comment
foster.jordan25d ago
Whoa, that's a rough one, but honestly good on you for catching it. It's wild how those little "rules" sneak in and we just accept them because they're already in the code, right? Like, I've seen the same thing with "stability" metrics, where someone's job hopping gets flagged, but nobody stops to think, maybe they moved for a partner's job or had to leave because of school. Once you actually break down why those gaps or changes matter, it's usually just a lazy way to filter for people who had a easy path, not actually better workers. The fact that you had to get a nudge from a friend to see it is the scariest part, because it means the rest of the team probably would have shipped it without a second thought.
9