Algorithmic systems in recruitment: how do players consider the risk of bias? Presentation of an empirical study serving a legal analysis

This paper presents a research in progress. The Neutr-IA project is financed by the David-Constant Fund and based at the University of Liège, Belgium. The Neutr-IA project proposes a legal interpretation of the discriminatory biases that emerge when artificial intelligence-based tools are used in recruitment, especially from a gender perspective. The research is based on a qualitative survey, that aims to assess whether the risks of bias have been considered, whether the behaviour of the recruitment services in question has been adapted, and whether the relevant legislation has been complied with. By cross-referencing the views of human resources departments, legal departments, and union representatives from the same company, we aim to identify the convergences, divergences and paradoxes emerging from these discourses. Based on the empirical results, we will analyse whether the legislation currently in force, and those currently being drafted, can be used to thwart, or even correct, the pitfalls of these new practices. The aim of this communication is not to present results - which are not yet available - but to outline the research objectives and the content of the survey, and to give some initial orientations based on the interviews conducted.