Bias in recruitment is persistent, unconscious and often already present before a candidate has even been spoken to. According to French research, as much as 85% of all bias arises before the first interview, particularly during CV selection. So it is not as though this first crept into the recruitment process with the arrival of A.I.; quite the opposite. But A.I. has recently put greater emphasis on it once again. Will A.I. only worsen this kind of bias and employment discrimination because it is trained on biased data? Or can it instead counter unjustified human prejudices in recruitment?
Will A.I. only worsen bias because it is trained on biased data?
A new white paper by Digitaal-Werven, sponsored by Selection Lab, Textmetrics, SoftMatch and TA Audit Institute, attempts to answer that question. It shows, for example, that if you only let A.I. look at CVs, much of the bias simply remains. The predictive value of a CV is, after all, low, as has often been shown. And very little changes if you look at it not with human eyes, but with an algorithm. So letting A.I. look at the whole CV and then training it on hires will always be biased. And that is what emerges every time, reports Bas van de Haterd, author of the white paper.
Read this article in full
Become a Werf& member for unlimited access to all articles, research and exclusive insights into the labour market.
No obligations Cancel anytime Secure payment