KI and the Labor Market: An Analysis of Employment Effects

Currently, the focus is on artificial intelligence applications, which have come to the attention of the general public through chatbots such as ChatGPT or Google Bard. In the first part of this study, the scientific studies on the labour market effects of AI are presented and the strengths and weaknesses of three different approaches are discussed. These approach the research question by means of i) determining how employees are potentially affected, ii) evaluating job advertisements and iii) using survey data on the use of AI applications. In the second part of this study, the relationship between the use of AI and the actual and expected employee development of companies in Germany is examined empirically. For this purpose, several waves of the IW-Zukunftspanel, a regular company survey, are evaluated in the period between 2019 and 2022 with around 4,300 observations from over 2,800 companies.

The empirical analysis based on the IW-Zukunftspanel does not show a robust correlation between the use of AI procedures and the past as well as the expected future employee development at the company level. The changing signs of the AI coefficient could indicate opposing impact mechanisms via positive productivity effects and substitution effects of human labour, as discussed in the literature. In order to entangle these effects, however, more information would be needed on the type of AI applications in the companies. In summary, the scientific findings in the literature so far do not point to strong negative employment effects. Rather, the evidence suggests that AI has so far largely complemented human labour. However, technological advances of AI processes at the current edge are not yet reflected in the studies and thus research on the employment effects of AI is only at the beginning.