As well as other business functions, human resources are nowadays impacted by new technologies with the aim of facilitating and improving their work. As every new technology, the advantages are unfortunately linked to potential threats of the technology. The biases that influence many HR decisions, with a particular focus on gender biases and race biases are unfortunately detectable when applying machine learning algorithms.
In the present report, starting from a general overview about biases in the recruitment process, why and how they exist, we study gender and ethnic bias in machine learning. Then, we explore how these two kinds of biases negatively influence the digitized recruitment process, presenting a framework of the main criticalities detected.
It is evident from our results, that there exists a huge space for improvement on the detection of gender and etnicity biases conjuctly.