Artificial intelligence and unemployment: New insights

This paper investigates the impact of artificial intelligence on unemployment in the most high-tech and developed countries, using a theoretical model that is also supported empirically. The empirical methodology follows a nonlinear approach by using panel threshold and GMM-system estimations. The dataset covers the period 1998-2016, and includes 23 countries. The main results show that artificial intelligence has a nonlinear impact on unemployment, with the acceleration of the use of artificial intelligence reducing unemployment, but only occurring at low levels of inflation. In this case, no switch effect between displacement effect and replacement effect is registered Otherwise, the contribution of artificial intelligence to unemployment is rather neutral. (C) 2021 Economic Society of Australia, Queensland. Published by Elsevier B.V. All rights reserved.

Automated summary

The study reveals that the impact of artificial intelligence on unemployment is nonlinear, with the acceleration of AI reducing unemployment only at low levels of inflation. Otherwise, the contribution of AI to unemployment is rather neutral.