Read a study showing a hiring algorithm in Chicago was 40% less likely to recommend women for tech roles

I found it in a research paper from a university lab last year. Does this mean we need stricter audits, or is the bias just reflecting real-world hiring patterns that already exist?

3 comments

3 Comments

craig.alex4d ago

That Chicago study is a perfect example of garbage in, garbage out. If the algorithm was trained on ten years of resumes from a male-dominated field, it basically learned that "good candidate" equals "looks like the guys we hired before." So stricter audits would just confirm the machine is a really fast, really biased copycat. The fix isn't just checking the math more, it's forcing a rewrite of the rules it learned from, maybe by hiding names and schools from the training data. Otherwise we're just building a faster horse and buggy for the same old biased road.

the_thomas3d ago

Yeah, that "faster horse and buggy" line really hits home.

max_brown4d ago

Ask if the study checked whether the algorithm was trained on the company's own past hiring data. Because if it was, then yeah, it's just copying old human bias and calling it math (which is a huge problem). Stricter audits might just catch that it's working as designed, to copy bad patterns. The real fix is probably changing the data it learns from, not just checking the results more often.