Ai and Jobs: Evidence from Online Vacancies

Abstract

We study the impact of AI on labor markets, using establishment level data on vacancies with detailed occupational information comprising the near-universe of online vacancies in the US from 2010 onwards. We classify establishments as "AI exposed" when their workers engage in tasks that are compatible with current AI capabilities. We document rapid growth in AI related vacancies over 2010-2018 that is not limited to the Professional and Business Services and Information Technology sectors and is significantly greater in AI-exposed establishments. AIexposed establishments are differentially eliminating vacancy postings that list a range of previously-posted skills while simultaneously posting skill requirements that were not previously listed. Establishment-level estimates suggest that AI-exposed establishments are reducing hiring in non-AI positions even as they expand AI hiring. However, we find no discernible impact of AI exposure on employment or wages at the occupation or industry level, implying that AI is currently substituting for humans in a subset of tasks but it is not yet having detectable aggregate labor market consequences.

occupation2vec: A general approach to representing occupations

“…In contrast, our measure does not align well with the one of Brynjolfsson et al (2018), which seems to be picking up something different. The fact that the three established measures are picking up different aspects of AI is already found by Acemoglu et al (2020), who also find that the measures of Felten et al (2018) and Webb (2019) tend to coincide and be different than the one of Brynjolfsson et al (2018).…”

“…We find that our estimates of AI exposure generally behave as a mix of the measures of Felten et al (2018) and Webb (2019) with no clear relationship to that of Brynjolfsson et al (2018). The close similarity of Felten et al (2018) and Webb (2019) in contrast to Brynjolfsson et al (2018) was first noted by Acemoglu et al (2020). Occupations that have a high degree of AI have similar profiles as those that score high on abstract tasks.…”

Artificial intelligence and work: a critical review of recent research from the social sciences

“…The example of labour economics is informative. To assess the trajectories of labour markets, labour economists and computer experts first make lists of skills and abilities associated to particular occupations, which they gather from the datasets of national labour offices and other organisations (typically the Burning Glass Labor Insight or the O*Net in the US), or from online job vacancy listings (Acemoglu et al 2020). They then use statistical tools to connect these lists with what AI processes are assumed to be able to perform.…”

“…A study by Acemoglu, Autor, Hazell, and Restrepo (2020) concludes no significant effect yet in the last decade at the occupation or industry level in the US. This confirms Autor's work of the early 2010s, in which he cautioned against overly pessimistic conclusions about technological unemployment (Autor and Handel 2013;Autor et al 2015) The standard approach to discuss the likelihood of technological unemployment is by labour economists who devise methods to extrapolate the impact of technology on particular tasks and deduce from it the impact on occupations and industries more generally.…”

“…To take one example, the Joint Artificial Intelligence Centre of the US Department of Defense presents its initiative as based on "solutions that are aligned with America's laws and values". There is a promise in the statement that AI innovations will deliver the kind of "good AI", or "AI for good", progressive-minded researchers and citizens are hoping for (Acemoglu 2021). But there is a more antagonistic aspect to such declarations, namely that the values embedded in the ethics of US-led AI will be American ethics, that is, a particularly American way of interpreting the core ethical norms that are to drive AI programs and algorithms.…”