THE FUTURE ROLE OF MACHINE LEARNING IN HR DEVELOPMENT

Using a Gaussian process classifier, Frey and Osborne (2017) examine how jobs are susceptible to computerisation in the US labour market by estimating the probability of technological replacement for 702 detailed occupations. The non-neutrality of technology and computerisation in labour markets, in general, is acknowledged in the literature (for a survey, see Piva & Vivarelli, 2017). Indeed, an increasing number of empirical studies (Marcolin et al., 2016) documents the decline of employment in routine intensive jobs, i.e., jobs mainly consisting of tasks following well-defined procedures that can be easily performed by sophisticated algorithms. Our research focuses on a specific occupation, the human resources (HR) manager, which is positioned at the 28th place out of the 702 occupations in the Frey and Osborne study. This evidence suggests that this occupation is hardly replaceable by technology (Chui et al., 2015). However, by disentangling the different tasks included in this occupation, we mainly focus on a complex and non-routine lever in human resource management (HRM) that requires intuition and discretionality, i.e. potential evaluation. By using a precautionary approach, this work tries to understand if machine learning can be effectively used in executing potential evaluation and consequently whether it might assist or even replace HR managers