Beyond BLS briefly summarizes articles, reports, working papers, and other works published outside BLS on broad topics of interest to MLR readers.
Historically, new technologies have driven major transformations in production processes and labor markets, often displacing low- and medium-skilled human labor. In recent years, an impending revolution in artificial intelligence (AI) has raised the specter of similar impacts, although this time around many researchers have focused on how AI can automate jobs at the high end of the skills distribution. At the same time, comparatively less effort has been devoted to studying how new AI technologies, by virtue of their importance for product innovation, can affect the internal organization and labor force composition of firms.
Consistent with this theme, in “Firm investments in artificial intelligence technologies and changes in workforce composition” (National Bureau of Economic Research, Working Paper 31325, June 2023), Tania Babina, Anastassia Fedyk, Alex X. He, and James Hodson offer new insights into how greater firm-level investment in AI can drive changes in firm organization and labor composition. To carry out their study, the authors construct a matched employer–employee dataset, drawing information from job postings and worker resumes. Firm investment in AI is measured with the intensity of hiring AI-skilled labor, whereas workforce composition and organization are gauged by data on worker education, specialization (major), and seniority.
With respect to organizational structure, the authors find that greater investment in AI tends to have a “flattening” effect on firm structure, whereby firms grow their workforce of entry-level employees and reduce the number of mid- and senior-level management positions. As illustrated in the paper, an increase of
Greater firm investment in AI also appears to have substantial labor composition effects. Most notably, such investment leads to “upskilling,” with firms increasing the share of their workers with undergraduate and graduate degrees and decreasing the share of those without college education. As one example, the authors report that, as AI hiring intensity rises by 1 standard deviation, the share of workers with undergraduate degrees jumps by 3.7 percent and the share of workers without college education drops by roughly twice that percentage. This result is also consistent with data on experience and educational requirements gleaned from job postings.
Finally, and also related to labor composition, AI investment tends to be associated with an increasing share of firm employees with degrees in STEM disciplines and a decreasing share of workers with social science degrees. Job postings data also reveal an employer preference for people with skills and experience in information technology and data analysis. Overall, the authors conclude that their findings challenge the view that AI poses a displacement risk to high-skilled labor, although they note that its impacts involve skilled-labor reallocation (toward AI-investing firms) and remain unclear at the aggregate (as opposed to firm) labor market level.