The Impact of Artificial Intelligence on the Labor Market

By Michael Webb. 2019

463 citations recorded by the source corpus.

I develop a new method to predict the impacts of a technology on occupations. I use the overlap between the text of job task descriptions and the text of patents to construct a measure of the exposure of tasks to automation. I first apply the method to historical cases such as software and industrial robots. I establish that occupations I measure as highly exposed to previous automation technologies saw declines in employment and wages over the relevant periods. I use the fitted parameters from the case studies to predict the impacts of artificial intelligence. I find that, in contrast to software and robots, AI is directed at high-skilled tasks. Under the assumption that the historical pattern of long-run substitution will continue, I estimate that AI will reduce 90:10 wage inequality, but will not affect the top 1%.

Subfields: Economics and Econometrics, Sociology and Political Science, Economics & Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

In Sociology and Political Science

In Economics & Operations Research

Citations in the corpus

Llama 2: Open Foundation and Fine-Tuned Chat Models (2023, background): “Lastly, broader societal issues like job displacement due to accelerated AI research and an over-reliance on LLMs leading to training data degradation are also pertinent considerations (Acemoglu and Restrepo, 2018; Autor and Salomons, 2018; Webb, 2019; Shumailov et al., 2023).”
Ethical and social risks of harm from Language Models (2021, background): “Unemployment and wages If LM-based applications displace employees from their roles, this could potentially lead to an increase in unemployment (Acemoglu and Restrepo, 2018; Webb, 2019), and other longer-term effects.”
GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models (2023, methodology): “Coeﬃcients on AI scores from (Webb, 2020) are also positive and statistically signiﬁcant at a 5% level, but our secondary prompt on overall exposure to LLMs in columns 3 and 4 does not exhibit a statistically signiﬁcant relationship.”
Taxonomy of Risks posed by Language Models (2022, background): “Advances in LMs and the language technologies based on them could lead to the automation of tasks that are currently done by paid human workers, such as responding to customer-service queries, with negative effects on employment [3, 192].”
Advanced Technologies Adoption and Use by U.S. Firms: Evidence from the Annual Business Survey (2020, unknown): “For instance, digitized information on workers and job postings have been used to track the supply and demand of IT labor as a proxy for IT diffusion in the economy (e.g., Tambe and Hitt 2012, 2020; Webb 2019; Bana et al, 2020; Tambe et al. 2020; Atalay et al. 2020).”
Exposure of occupations to technologies of the fourth industrial revolution (2021, methodology): “, Webb (2019) used O*NET task descriptions from 2017.”
Occupational, industry, and geographic exposure to artificial intelligence: A novel dataset and its potential uses (2021, methodology): “Interestingly, Acemoglu et al. also do the same for the Brynjolfsson et al. (2018) and Webb (2020) measures and discuss the differences across the three measures.”
New Frontiers: The Origins and Content of New Work, 1940–2018 (2022, methodology): “Our procedure for measuring automation innovations follows closely on (Kogan et al., 2019; Webb, 2020), which use the textual similarity between occupational task content and patent texts to measure the ability of new technologies to perform the same work done by workers in particular occupations.”
The Labor Market Impacts of Technological Change: From Unbridled Enthusiasm to Qualified Optimism to Vast Uncertainty (2022, methodology): “Drawing on AI-suitability indexes developed by Brynjolfsson et al. (2018), Felten et al. (2018), and Webb (2020), the paper first predicts which establishments are likely to adopt AI as a function of the suitability of their job task structures (visible in job postings) in the pre-AI era.”
Tasks, Automation, and the Rise in US Wage Inequality (2021, methodology): “Most importantly, the results are similar when we rely on the measure of automatable jobs constructed by Webb (2020) based on the text of new patents.”
Digital technologies, innovation, and skills: emerging trajectories and challenges (2021, methodology): “By connecting AI patent descriptions to the occupational descriptions in O*NET, Webb (2019) uses a slightly different strategy to connect AI to tasks and finds that, in contrast to earlier rounds of automation related to software and robots, AI may be able perform tasks that previously were…”
Going digital: implications for firm value and performance (2019, unknown): “…of AI technology exposure at the industry-level, which is estimated by the cosine similarity between the abstracts/titles of AI-related patents (which are identified using the methodology of Webb 2020) and NAICS industry descriptions (following the GloVe vector approach of Kogan et al. 2022).”
How will Language Modelers like ChatGPT Affect Occupations and Industries? (2023, background): “Second, the article builds on and extends a set of papers that provide systematic methodologies for studying how AI affects occupations (e.g., Brynjolfsson et al, 2018; Frey & Osborne, 2017; Tolan et al., 2021; Webb, 2020).”
Artificial intelligence, firm growth, and product innovation (2024, background): “OAGv2 provides a unified view of two large-scale databases of academic paper metadata, abstracts, citations, and author links: (i) the Microsoft Academic Graph (part of the Microsoft Academic Service infrastructure in Sinha et al. (2015)), and (ii) ArnetMiner (Tang et al.”
Measuring the Occupational Impact of AI: Tasks, Cognitive Abilities and AI Benchmarks (2020, background): “In a more recent article, Webb (2020) measures AI's occupational impact by computing the overlap be-tween O*NET job task descriptions and the text of patents.”
AI and Jobs: Evidence from Online Vacancies (2020, background): “measures: Felten, Raj and Seamans’ (2018, 2019) AI Occupational Impact measure; Brynjolfsson, Mitchell and Rock’s (2018, 2019) Suitability for Machine Learning (SML) index; and Webb’s (2020) AI Exposure score.”