How GoRecruit deals with the «black box» problem

One of the major challenges of artificial intelligence is the «black box» problem — if we do not know how AI works, we will not trust it. This is particularly an issue when artificial intelligence makes responsible decisions and not just in the case of such spheres as medicine, but also HR. Because when introducing AI in the candidate selection process one has to keep in mind that unjustified denial of employment is grounds for court action. Let us tell you to what extent the «black box» problem is relevant to GoRecruit and how we handle it!

In fact, we have already dealt with the heart of the matter — the user’s incomprehension of the AI’s inner workings, because we explained how GoRecruit works. Here is a quick recap: the system is based on the analysis of the data from open sources and social networks, the main method for processing profiles is text analysis. The service uses 16 socionic types to evaluate the way a person behaves himself in open sources. To identify the type GoRecruit needs three parameters, two of them is the result of a text analysis, we get the third one by analyzing other open sources.

The type of the person coupled with other data is loaded into the mathematical model, which in turn does not make a decision on whether a job applicant is a good fit for a vacant position, but assigns a rating and gives the possibility for a candidate comparison. So that an HR specialist can save time and instead of conducting tens of interviews, invite only the job applicants with the highest scores. GoRecruit supports a recruiter and always leaves the ultimate decision to him.