Loading…
9
I was sure the hiring algorithm my company bought was fine until I watched it screen out 47 people with one typo
I sat in on a demo of this resume screener we were testing at my office in Columbus back in March and the sales guy kept saying how neutral it was, no humans, no bias. Then I watched it dump 47 applicants from one zip code into the reject pile in about 90 seconds, and the hiring manager just shrugged and moved on to lunch. I asked to see the training data and nobody could tell me where it came from, so I pushed for two weeks to get a human review step added before any auto reject. It took three meetings and a lot of pushback, but we now manually check anyone the system flags, and the first pool pulled in 12 people we would have never seen. Has anyone else had luck getting their company to slow down a tool like this, or did you just get told to trust the math?
1 comments
Log in to join the discussion
Log In1 Comment
irisowens1mo ago
How many of those 47 were actually good fits though? A zip code filter sounds bad until you realize some zip codes are just far from the office, and resume screeners mostly sort on stuff humans would skip anyway. I get wanting a second look, but one demo with 47 rejects is not proof the whole thing is broken, it might just mean the tool is doing the boring part faster than you can watch.
4