Artificial intelligence and machine learning in the world of recruiting

Artificial Intelligence and Machine Learning applied to the HR sector are topics becoming more and more popular nowadays. However, we don’t often know enough about it to be able to fully understand what we are told.

Artificial Intelligence is a technology capable of rapidly processing a large amount of data (reading and sorting). AI was born in response to a concrete problem that the human being is no longer able to face in the traditional way: “big data management”.

Parameterizing expectations is primary. We cannot think that technology is currently capable of conducting a complete analysis of any usable content and making discretionary decisions, prerogatives that remain of the human being. Artificial Intelligence carries out support activities; in most cases these are “fatigue tasks”, linked to a theme of quantity, repetitive and with low added value.

In the world of recruiting we have – and will increasingly continue having – a problem related to reading CVs. Let’s see two concrete examples together:

• Often, the high CV volume received do not allow recruiters to analyze all CVs – this increases the risk of losing profiles with important skills;
• Professions are increasingly “fluider” and therefore forcing the inclusion of some profiles within specific roles, risks making us miss out on opportunities.

The Artificial Intelligence linked to the world of recruiting allows you to read all incoming CVs and make a semantic analysis on them, determining a matching between offers and applications, and a ranking of the profiles selected according to their compatibility with the search.
In this way, the recruiter avoids mechanical and repetitive work, and can work qualitatively on the pre-screening of the CVs provided by the machine, with a significant saving in terms of time and resources.

The answer is no. The semantic reading engines that underlie the AI applied to the Talent Acquisition sector perform different actions and have different values. It is our job to test them and evaluate their effectiveness.
The largest semantic and artificial intelligence engines are governed by the world’s largest “tech” operators: Google, Amazon AWS, IBM, Microsoft. Most of their services are free, or in any case easily accessible to our HR software. The problem is that these giants are by definition not verticalized linguistic engines. This means that the article in a newspaper and / or a novel is read correctly but vertical documents such as financial reports or medical records – which do not use “classic” writing – are not read effectively.
The same applies to a CV (also a vertical document); for a complete reading it is not enough to have a semantic linguistic engine that binds the concepts but you need an Artificial Intelligence developed specifically for the world of work.

The example of “black gold” is often used to show how the machine, independently, is able to trace the expression back to the term “oil”. If these conceptual links work in daily writing and speech, they do not find an effective application in the world of human resources; it is not difficult to believe that no oil & gas operator will write in his CV “black gold”.

Another note of attention is when an AI is built specifically for the world of HR, but are used institutional mappings. The European cataloging, and those that every single state does, have little functional impact for the real writing of CVs and job offers (it is too different and not comparable). It is functional for statistical reports but less so for real use during a Recruitment process.

On the market there are HR TECH operators offering ATS equipped with Artificial Intelligence, which, however, are often not vertical on the real needs of customers. Although these are new technologies and semantic engines, few systems are built precisely for the HR sector (and therefore aimed at recruiting and, more generally, Talent Acquisition).