Artificial Intelligence and Traditional Machine Learning to Deep Neural Networks: A Study for Social Implications

Abstract

There are intriguing opportunities on the road to broad usage of artificial intelligence, as well as challenges that must be solved in order to use machine learning and artificial intelligence technology in industries. On the basis of Google Scholar data, this work systematizes artificial intelligence sections and analyzes the dynamics of changes in the number of scientific papers in machine learning sections. The process of data collection and calculation of dynamic indicators of changes in publication activity, such as the growth rate and acceleration of growth of scientific publications, is presented in detail. Particularly evident in the analysis of publishing activity was the high level of interest in current transformer models, the production of datasets for specific industries, and a significant growth in interest in explainable machine learning approaches. As indicated by the negative correlation between the number of publications published and the number of articles published in a given year, relatively minor study topics are gaining an increasing amount of attention. Results demonstrate that, despite the method's limitations, it is possible to (1) identify rapidly expanding fields of research, irrespective of the number of papers published, and (2) anticipate publication activity in the short term with sufficient accuracy for practical purposes. Results for more than 400 search queries relating to designated study topics and the application of machine learning models to industries are presented in this article in a single document. The suggested method assesses the growth and fall of scientific areas that are related with specific key phrases in terms of their dynamics of growth and decline. It does not necessitate access to huge bibliometric archives and enables for the generation of quantitative estimates of dynamic indicators in a reasonably short time frame.