Thesis Public Defense | VNP26 - Nguyễn Phú Quốc

Artificial intelligence and unemployment: An international evidence

Student: Nguyễn Phú Quốc, VNP 26

Supervisor: Dr. Võ Hồng Đức

Abstract:

The term artificial intelligence (AI) was first defined in 1956 by John McCarthy as “a machine that behaves in ways that would be called intelligent if a human were so behaving” or “the science and engineering of creating intelligent machines”. The initial study and its modest achievements of AI with mere pre-programmed performances remained far beyond human intellectual abilities, though the discipline functioned as a very foundation for further development in the field.

Nowadays, modern information technologies and the advent of cognitive machines powered by AI have powerfully transformed people’s life and work. While AI can amplify productivity in some industries, there are instances where they can take over human work and revolutionize occupations to some degree. This paper scrutinizes the possible influence of AI on unemployment using a broad database of AI-related patents in 40 developed and emerging markets from 2000 to 2019. The study employs a panel smooth transition regression (PSTR) model to analyze the relationship between AI and unemployment under various inflation levels. The study contributes to the existing literature with several findings. First, findings from our analysis confirm the non-linear relationship between AI
and unemployment depending on the threshold of inflation. In general, AI increases unemployment until a certain threshold of inflation is attained, then the disclosed ascendancy reduces its effect afterwards. Second, the smooth mechanism employed in this analysis can capture individual estimates varying among countries over time. Finally, this pioneering research extends the literature coverage to provide evidence on both developed and emerging economies.

Keywords: Artificial intelligence (AI), patents, unemployment rate, panel smooth threshold regression (PSTR)