As a recruiter, you have almost certainly used software to organize CVs according to their relevance, saving you hours of troublesome screening. You simply define the profile with specific criteria—for example, a master’s degree and fluency in English—and the software takes it from there. The algorithm selects CVs with complete objectivity, based on the rules you choose.
Bear in mind that humans make the decisions using specific filter criteria, not the machine. When it comes to the preliminary sorting of CVs, the notion that a human does better than a machine is difficult to justify. When you are facing a huge number of CVs, a filter’s functionality can no longer be challenged. Indeed, it permits human resources to focus on high-value tasks and potential employees’ more human qualities.
But what if an intelligent machine could go beyond your preselection criteria, sorting candidates according to their turnover probability, for example? In this way, it would create its own filters in order to meet your requests.
The machine creates its own criteria in a way that is more than objective. Using machine learning, the algorithm concludes which candidates are most likely to quickly leave the company based on the information contained in their CVs.
It may sound futuristic, but this procedure is currently being tested in India, where turnover is a substantial issue in certain big enterprises.
How does it work? Using past information, the software can identify recurrences and find correlations between employees who leave directly after hiring and the contents of the CV. It then calculates the predictive probability of turnover of each CV based on its contents.
In a world in which information and personal facts are more and more distributed, it is easy to see how machine learning in the service of artificial intelligence (AI) can be modulated and applied in many recruitment situations. AI offers some interesting opportunities to recruiters, including saving time and gaining productivity, as shown in the above example.
One developer who works with predictive models of AI within social networks in the domain of information security estimates that such models are 80% accurate. However, he remains skeptical when it comes to AI in human resources. The subtleties of communication in a CV are more complex as they are human. As the workplace becomes increasingly globalized and culturally intertwined, should we consider using science in place of human communications?
As a counterargument, one of the best collaborators on one particular team also submitted the worst CV. He would never have been selected by a machine because machines do not make exceptions. A machine can evaluate only what is compliant with what it has learned. For a tech company looking for informatics talent, the process of selection by an algorithm is absurd. The question of whether a candidate’s CV conforms to particular guidelines is not necessarily what determines the strongest candidate.
So is conformity good for all enterprises? Consider this perspective: Why should a candidate lose the possibility of access to a human recruiter simply because their CV is not keyword-compliant?
Legally, how can we justify the choices a machine makes? And what about the experience of the machine—does it need to be regulated? Due to the underlying process of machine learning, every machine has a sample whose width is scalable, which means each one judges a CV differently.
Recruiters must consider these deliberations and use AI as a tool that can be applied in a productive way that is specific to the company, the sector, and the work environment. AI must remain a tool to be utilized in a reasonable way that does not call the law into question.
For more on future-focused hiring strategies, see “Which Employees Do We Need In The Future? Not Only The Masterminds.”