Office notebook. When a recruiter is faced with two opposing recommendations for hiring the new kid, one from a human and the other from artificial intelligence (AI), which one will he follow?
Two researchers in management sciences, Alain Lacroux, from the University of Paris-I Panthéon-Sorbonne, and Christelle Martin-Lacroux, from the University of Grenoble-Alpes, were interested in the question, to see more clearly the influence of algorithms on the selection of application files. The conclusions of their experimental studies, carried out as a preamble to a call for research projects for the Ministry of Labor and published at the end of September, demonstrate that you can never distrust yourself enough.
The use of artificial intelligence in human resources, relatively recent in France, has reached companies: “50% of recruiters say they use at least one AI-related tool during their process”, says Christelle Martin-Lacroux.
At each stage of recruitment – sourcing, drafting job offers, application qualification, interview, matching, etc. –, artificial intelligence is a promise of efficiency gained: diversification of applications, non-biased announcements, rapid prequalification, deciphering of non-verbal language during interviews, more objective choice through algorithmic matching, etc. But what do human resources managers actually do with it?
Even when he’s wrong
The two researchers’ experiment consisted of evaluating the reactions of a panel of recruiters to two CVs submitted to them, one relevant, the other not, accompanied by a recommendation made either by an algorithm or by an HR manager. First, the recruiter had to express their level of confidence in the recommendation, before rating the CV, then choosing one.
During the experiment, recruiters said they had more confidence in the human expert than in the algorithms, even when he made a mistake and recommended the worst candidate. It is true that there is no shortage of reasons to be wary of the assistance of artificial intelligence, such as discriminatory biases for example.
However, when choosing the CV, to the great astonishment of researchers, and despite declarations of greater confidence in the recommendations of human resources managers, recruiters followed the advice of algorithms. “Conflicting recommendations can lead recruiters to a form of cognitive lazinesssays Alain Lacroux, which is based on the following logic: since I don’t know how it works but it is recommended to me, it must not be that bad.
You have 10% of this article left to read. The rest is reserved for subscribers.