STORY · FORSKNING_
AI systems develop their own biases in hiring processes
New research shows that language models not only inherit bias from training data, but also develop their own skewness when used to screen job applicants. This happens independently of bias in the original data, making the problem more complicated than previously assumed.
WHY IT MATTERS
Since AI systems are increasingly used for initial screening of CVs, it is critical to understand that these systems can discriminate in ways that are not solely due to human historical bias – something that makes it harder to address the problem through training data cleaning alone.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.