A two-stage procedure for optimal modeling of the probability of training needs in artificial intelligence

Nowadays artificial intelligence (AI) has shown a rapidly increasing role in supporting or improving decision-making. Many researchers have studied and applied AI-systems in several fields of research. However, few studies have assessed the effects of AI applications on training needs. This paper proposes an innovative two-stage analysis to investigate Awareness, Attitude and Trust towards AI and their reflections on learning needs. In particular, it is shown how a machine learning variable selection algorithm can support an optimal multilevel modeling. First of all, the Boruta Random Forest algorithm is used as a variables selection method to define the optimal subset of all relevant variables; then these variables are included as covariates of a multilevel binary model needs which takes into account the variability among different countries for estimating the probability of educational. Indeed, in this cross-sectional study, addressing citizen’s perspectives with respect to this topic is crucial to gain a better understanding of people’s views and perceptions regarding the relative complexity of AI that can breed fear and mistrust as well as new needs in modern living European context. For this analysis a dataset obtained from a survey regarding AI and ethics on European citizens distributed in eight countries is considered. This repository contains data generated for the manuscript: " A two-stage procedure for optimal modeling of the probability of training needs in artificial intelligence". It comprehends: (1) the dataset Data_Boruta_Random_Forest used to estimate the variables importance. (2) the dataset Data_Multilevel to perform the comparison among different multilevel binary models proposed in the paper.